Mold instantaneous liquid level control method and device, storage medium and computer equipment

Abstract

The application discloses a crystallizer instantaneous liquid level control method and device, a storage medium and a computer device. The method comprises the following steps: collecting an instantaneous liquid level value of molten steel; predicting an instantaneous liquid level prediction value of the molten steel based on a preset liquid level prediction model and the collected instantaneous liquid level value; whenever it is determined that the liquid level state of the molten steel in the crystallizer is abnormal based on the predicted instantaneous liquid level prediction value, predicting a stopper height prediction set value based on a stopper height prediction model and the instantaneous liquid level prediction value, so that the hydraulic device adjusts the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer, in which the molten steel flow changes, is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value. The two prediction models are combined to obtain a cycle prediction model, so that the liquid level fluctuation stability in the production process is improved, and the production efficiency is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of instantaneous liquid surface abnormal fluctuation, in particular to a crystallizer instantaneous liquid surface control method and device, a storage medium and a computer device. BACKGROUND

[0002] As a core link of modern steel production, continuous casting has an irreplaceable position and role. However, due to the complexity of the continuous casting process, there are many surface defects in the continuous casting billet, for example, the liquid surface fluctuation of the crystallizer has a great influence on the quality of the continuous casting billet, and the abnormal fluctuation of the liquid surface of the crystallizer can greatly increase the probability of surface cracks of the continuous casting billet. Since the instantaneous abnormal fluctuation of the liquid surface of the crystallizer frequently occurs in high-speed production, the continuous casting production efficiency is greatly limited. SUMMARY

[0003] Therefore, the application provides a crystallizer instantaneous liquid surface control method and device, a storage medium and a computer device, which are applied to a continuous casting billet molten steel pouring system. The continuous casting billet molten steel pouring system sequentially includes a ladle, a tundish and a crystallizer along a molten steel pouring direction. A stopper is arranged in the tundish and can be adjusted in height by a hydraulic device. When the stopper is adjusted in height by the hydraulic device, the molten steel flow from the tundish to the crystallizer will change. The instantaneous liquid level value of the molten steel is collected in real time. Based on a preset liquid level prediction model and the instantaneous liquid level value, the instantaneous liquid level prediction value of the molten steel is predicted. When the liquid level state of the molten steel in the crystallizer is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value is predicted based on a stopper height prediction model and the instantaneous liquid level prediction value, so that the hydraulic device adjusts the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with the changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value. The two prediction models are combined to obtain a cycle prediction model, which can improve the liquid surface fluctuation stability in the production process.

[0004] According to an aspect of the application, a crystallizer instantaneous liquid surface control method is provided, which is applied to a continuous casting billet molten steel pouring system. The continuous casting billet molten steel pouring system sequentially includes a ladle, a tundish and a crystallizer along a molten steel pouring direction. The tundish is provided with a stopper which can be adjusted in height by a hydraulic device. When the stopper is adjusted in height by the hydraulic device, the molten steel flow from the tundish to the crystallizer will change. The method comprises:

[0005] collecting the instantaneous liquid level value of the molten steel in the crystallizer in real time;

[0006] predict, based on a preset liquid level prediction model and the instantaneous liquid level values collected in a preset historical collection period, an instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval, to obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure;

[0007] Whenever it is determined, based on the predicted instantaneous liquid level prediction value, that the liquid level state of the molten steel in the crystallizer at the future preset time interval is abnormal, a stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on a stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure;

[0008] The hydraulic device is called to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer in which the molten steel flow changes is re-predicted based on the preset liquid level prediction model, and it is determined, based on the re-predicted instantaneous liquid level prediction value, that the liquid level state of the molten steel in the crystallizer at the future preset time interval is normal.

[0009] Optionally, before the stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value, the method further comprises:

[0010] Historical data in the running process of the continuous casting billet molten steel pouring system is collected, wherein the historical data includes instantaneous liquid level historical values of the molten steel in the crystallizer and stopper height historical set values of the stopper in the tundish at different historical time points, and the instantaneous liquid level historical values include instantaneous liquid level historical values of the molten steel in the crystallizer when the liquid level state is normal and abnormal, respectively;

[0011] Based on the historical data after data preprocessing, a historical data training set, a historical data test set and a historical data validation set are obtained;

[0012] Based on the historical data training set, the historical data test set and the historical data validation set, a stopper height prediction model is trained.

[0013] Optionally, after the instantaneous liquid level value of the molten steel in the crystallizer is collected in real time, the method further comprises:

[0014] The collected instantaneous liquid level value is filtered and interpolated to obtain a to-be-predicted instantaneous liquid level value;

[0015] Correspondingly, the method comprises: predicting, at every preset time interval, an instantaneous liquid level value of the molten steel in the crystallizer at a future time interval based on a preset liquid level prediction model and an instantaneous liquid level value collected in a preset historical collection period, to obtain an instantaneous liquid level prediction value.

[0016] The method comprises: predicting, at every preset time interval, an instantaneous liquid level value of the molten steel in the crystallizer at a future time interval based on a preset liquid level prediction model and an instantaneous liquid level value collected in a preset historical collection period, to obtain an instantaneous liquid level prediction value.

[0017] Optionally, before the step of predicting the stopper height prediction set value that makes the abnormal liquid level state return to normal based on the stopper height prediction model and the instantaneous liquid level prediction value, the method further comprises:

[0018] performing fuzzy control on the instantaneous liquid level prediction value based on a fuzzy proportional-integral-derivative controller.

[0019] Correspondingly, the step of predicting the stopper height prediction set value that makes the abnormal liquid level state return to normal based on the stopper height prediction model and the instantaneous liquid level prediction value comprises:

[0020] predicting the stopper height prediction set value that makes the abnormal liquid level state return to normal based on the stopper height prediction model and the instantaneous liquid level prediction value after the fuzzy control.

[0021] Optionally, before the step of determining that the liquid level state of the molten steel in the crystallizer at the future time interval is abnormal based on the predicted instantaneous liquid level prediction value, the method further comprises:

[0022] determining the liquid level state of the molten steel in the crystallizer at the future time interval based on the instantaneous liquid level prediction value and a preset normal liquid level fluctuation range, wherein the liquid level state comprises abnormal and normal.

[0023] When the instantaneous liquid level prediction value is outside the preset normal liquid level fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be abnormal; and when the instantaneous liquid level prediction value is within the preset normal liquid level fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be normal.

[0024] Optionally, the method further comprises:

[0025] acquiring historical normal instantaneous liquid level values at a plurality of historical time points in a running process of the continuous casting billet molten steel pouring system, and determining a normal liquid level fluctuation range based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values comprise a plurality of different values.

[0026] Optionally, after the hydraulic device is called to readjust the height of the stopper in the tundish based on the predicted stopper height prediction setting value, the method further comprises:

[0027] The height of the stopper in the tundish is monitored in real time, and when it is monitored that the height of the stopper in the tundish after readjustment is not at the predicted stopper height prediction setting value, the height of the stopper in the tundish is readjusted by the hydraulic device and the stopper height prediction setting value again.

[0028] According to another aspect of the present application, a crystallizer instantaneous liquid level control device is provided, which is applied to a continuous casting billet liquid steel pouring system, the continuous casting billet liquid steel pouring system sequentially comprises a ladle, a tundish and a crystallizer along the liquid steel pouring direction, the tundish is provided with a stopper whose height can be adjusted by a hydraulic device, and when the height of the stopper is adjusted by the hydraulic device, the flow of liquid steel flowing from the tundish to the crystallizer will change, and the device comprises:

[0029] A field liquid level acquisition module is configured to acquire the instantaneous liquid level value of the liquid steel in the crystallizer in real time;

[0030] A future liquid level prediction module is configured to, at every preset time interval, predict the instantaneous liquid level value of the liquid steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and the instantaneous liquid level values acquired in a preset historical acquisition period, to obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure;

[0031] An abnormal liquid level stopper height prediction module is configured to, whenever it is determined that the liquid level state of the liquid steel in the crystallizer at a future preset time interval is abnormal based on the predicted instantaneous liquid level prediction value, predict a stopper height prediction setting value that makes the abnormal liquid level state return to normal based on a stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure;

[0032] A stopper height adjustment module is configured to call the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height prediction setting value, until the crystallizer whose liquid steel flow is changed based on the preset liquid level prediction model is predicted again, and it is determined that the liquid level state of the liquid steel in the crystallizer at a future preset time interval is normal based on the predicted instantaneous liquid level prediction value.

[0033] Optionally, the device further comprises a stopper height prediction model training module configured to:

[0034] collect historical data in a running process of a continuous casting billet liquid pouring system, wherein the historical data comprises different historical time, a historical value of an instantaneous liquid level of the steel liquid in a crystallizer and a historical set value of a stopper height of a stopper in a tundish, the historical value of the instantaneous liquid level comprises historical values of the instantaneous liquid level when the liquid level state of the steel liquid in the crystallizer is normal and abnormal respectively;

[0035] based on the historical data after data preprocessing, obtain a historical data training set, a historical data test set and a historical data validation set;

[0036] based on the historical data training set, the historical data test set and the historical data validation set, train a stopper height prediction model.

[0037] Optionally, the device further comprises a data preprocessing module for:

[0038] filtering and interpolating the collected instantaneous liquid level value to obtain a to-be-predicted instantaneous liquid level value.

[0039] Optionally, the future liquid level prediction module is further configured to:

[0040] based on a preset liquid level prediction model and the to-be-predicted instantaneous liquid level value collected in a historical preset collection period, predict an instantaneous liquid level value of the steel liquid in the crystallizer at a future preset time interval, to obtain an instantaneous liquid level prediction value.

[0041] Optionally, the device further comprises a fuzzy control module for:

[0042] based on a fuzzy proportional-integral-derivative controller, perform fuzzy control on the instantaneous liquid level prediction value.

[0043] Optionally, the abnormal liquid level stopper height prediction module is further configured to:

[0044] based on the stopper height prediction model and the fuzzy-controlled instantaneous liquid level prediction value, predict a stopper height prediction set value that makes the abnormal liquid level state return to normal.

[0045] Optionally, the device further comprises an abnormal liquid level determination module for:

[0046] based on the instantaneous liquid level prediction value and a preset liquid level normal fluctuation range, determine a liquid level state of the steel liquid in the crystallizer at the future preset time interval, wherein the liquid level state comprises abnormal and normal;

[0047] when the instantaneous liquid level prediction value is outside the preset liquid level normal fluctuation range, the liquid level state of the steel liquid in the crystallizer is determined to be abnormal, and when the instantaneous liquid level prediction value is within the preset liquid level normal fluctuation range, the liquid level state of the steel liquid in the crystallizer is determined to be normal.

[0048] Optionally, the abnormal liquid level determination module is further configured to:

[0049] acquire historical normal instantaneous liquid level values at a plurality of historical moments during operation of the continuous casting billet molten steel pouring system, and determine a normal fluctuation range of the liquid level based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values include a plurality of different values.

[0050] Optionally, the device further comprises a stopper height monitoring module configured to:

[0051] acquire historical normal instantaneous liquid level values at a plurality of historical moments during operation of the continuous casting billet molten steel pouring system, and determine a normal fluctuation range of the liquid level based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values include a plurality of different values.

[0052] According to yet another aspect of the present application, a storage medium having a computer program stored thereon is provided, the program being executed by a processor to implement the above-mentioned mold instantaneous liquid level control method.

[0053] According to still another aspect of the present application, a computer device is provided, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, the processor implementing the above-mentioned mold instantaneous liquid level control method when executing the program.

[0054] Through the above technical solutions, the present application provides a mold instantaneous liquid level control method and device, a storage medium, and a computer device, which acquire an instantaneous liquid level value of molten steel in real time, predict an instantaneous liquid level prediction value of the molten steel based on a preset liquid level prediction model and the instantaneous liquid level value, predict a stopper height prediction set value based on a stopper height prediction model and the instantaneous liquid level prediction value whenever the liquid level state of the molten steel in the mold is determined to be abnormal based on the predicted instantaneous liquid level prediction value, so that the hydraulic device adjusts the height of the stopper in the tundish based on the predicted stopper height prediction set value until the mold in which the molten steel flow changes is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the mold is determined to be normal based on the re-predicted instantaneous liquid level prediction value. The combination of the two prediction models can improve the stability of liquid level fluctuation in the production process.

[0055] The above description is only a summary of the technical solutions of the present application, in order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented in accordance with the content of the description, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0056] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0057] Figure 1 A flow chart of a crystallizer instantaneous liquid level control method provided by an embodiment of the application is shown;

[0058] Figure 2 A continuous casting billet liquid steel pouring system provided by an embodiment of the application is shown;

[0059] Figure 3 A flow chart of another crystallizer instantaneous liquid level control method provided by an embodiment of the application is shown;

[0060] Figure 4 A fuzzy proportional-integral-derivative controller control architecture provided by an embodiment of the application is shown;

[0061] Figure 5 A flow chart of still another crystallizer instantaneous liquid level control method provided by an embodiment of the application is shown;

[0062] Figure 6 A structure diagram of a crystallizer instantaneous liquid level control device provided by an embodiment of the application is shown;

[0063] Figure 7 A structure diagram of another crystallizer instantaneous liquid level control device provided by an embodiment of the application is shown.

[0064] In the drawings, 21 is a ladle, 22 is a tundish, 23 is a hydraulic device, 24 is a liquid level probe, 25 is a crystallizer, 26 is liquid steel, and 27 is a stopper. DETAILED DESCRIPTION

[0065] The application will be described in detail below with reference to the drawings and embodiments. It should be noted that the embodiments in the application and the features in the embodiments can be combined with each other without conflict.

[0066] In the embodiment, a crystallizer instantaneous liquid level control method is provided, as shown in Figure 1 applied to a continuous casting billet liquid steel pouring system, the continuous casting billet liquid steel pouring system includes a ladle, a tundish and a crystallizer in sequence along a liquid steel pouring direction, the tundish is provided with a stopper with a height adjustable by a hydraulic device, and the liquid steel flow from the tundish to the crystallizer will change after the height of the stopper is adjusted by the hydraulic device, and the method includes:

[0067] In step 101, the instantaneous liquid level value of the liquid steel in the crystallizer is collected in real time.

[0068] In the above embodiments of the present application, the liquid level control device integrated into the continuous casting billet liquid pouring system can control the instantaneous liquid level of the crystallizer as a whole to achieve rapid and accurate control of the liquid level fluctuation of the molten steel during the control process. Specifically, the continuous casting billet liquid pouring system, for example Figure 2 As shown in the figure, the molten steel 26 first flows through the ladle 21, then flows into the tundish 22, and then flows into the crystallizer 25. In the tundish 22, a stopper 27 can be adjusted in height by a hydraulic device 23. When the stopper 27 is adjusted in height by the hydraulic device 23, the flow of the molten steel 26 from the tundish 22 to the crystallizer 25 will change. The crystallizer 25 is provided with a liquid level probe 24, which can be used to collect the instantaneous liquid level value of the molten steel in the crystallizer in real time. In particular, the instantaneous liquid level value of the molten steel can be collected by a 10Hz collection frequency.

[0069] Step 102: At every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted based on a preset liquid level prediction model and the instantaneous liquid level values collected in a preset historical collection period, to obtain an instantaneous liquid level prediction value. The preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0070] Then, every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval moment is predicted based on the preset liquid level prediction model and the instantaneous liquid level values collected in the preset historical collection period, to obtain an instantaneous liquid level prediction value. The preset time interval can be set to 2s, that is, the preset liquid level prediction model predicts the instantaneous liquid level value of the molten steel in the crystallizer at the second 2s in the future every 2s. In particular, the preset liquid level prediction model can be set to 1200 instantaneous liquid level values as input, and correspondingly the preset historical collection period can be set to 2 minutes. The preset liquid level prediction model can be established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure. The BI-LSTM network structure includes two independent LSTM network layers, which are referred to as a forward LSTM layer and a backward LSTM layer. The forward LSTM layer processes data from front to back in the natural order of sequence data, while the backward LSTM layer processes data from back to front in the opposite order. In the forward LSTM layer, each LSTM unit receives the output from the previous LSTM unit and the input at the current time step, and updates and filters information through mechanisms such as input gates, forget gates, and output gates, and finally outputs the hidden state at the current time step. Similarly, in the backward LSTM layer, each LSTM unit also receives the output from the previous LSTM unit (but at this time it is the previous unit in the backward order) and the input at the current time step, and performs similar information filtering and updating operations. In the output layer of the BI-LSTM network structure, the hidden states of the forward LSTM layer and the backward LSTM layer at the same time step are spliced to form the final hidden state at the time step. Therefore, the BI-LSTM network structure can capture the forward and backward context information of the sequence data at the same time, thereby more comprehensively understanding the overall structure of the sequence data. The BI-LSTM network structure can effectively solve the problem that the traditional LSTM network can only pass information in one direction when processing sequence data, and improve the ability to capture context information of sequence data. At the same time, the BI-LSTM network structure performs well in sequence data processing tasks such as natural language processing and speech recognition, and has wide application prospects. As for the training of the preset liquid level prediction model, the historical instantaneous liquid level values at multiple historical moments in the preset historical period and a target instantaneous liquid level value corresponding to the multiple historical instantaneous liquid level values can be obtained in advance as a training data set to train the preset liquid level prediction model, so that the preset liquid level prediction model predicts future liquid level values based on historical liquid level values.

[0071] Step 103, whenever the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0072] Then, whenever the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value, so that the compensation position of the stopper can be predicted in advance to prevent the occurrence of abnormal fluctuations.

[0073] Step 104, calling the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

[0074] Finally, calling the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value. Therefore, the two prediction models are combined to obtain a cycle prediction model, which can improve the stability of liquid level fluctuation in the production process.

[0075] By applying the technical solution of the embodiment, the instantaneous liquid level value of the molten steel is collected in real time; the instantaneous liquid level prediction value of the molten steel is predicted based on the preset liquid level prediction model and the instantaneous liquid level value; whenever the liquid level state of the molten steel in the crystallizer is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value, so that the hydraulic device readjusts the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value. By combining the two prediction models to obtain a cycle prediction model, the stability of liquid level fluctuation in the production process can be improved.

[0076] Further, as a refinement and extension of the above embodiment, in order to fully describe the specific implementation process of the embodiment, another mold instantaneous liquid level control method is provided, which is applied to a continuous casting billet liquid steel pouring system. The continuous casting billet liquid steel pouring system includes a ladle, a tundish and a mold in sequence along the liquid steel pouring direction. A stopper with adjustable height by a hydraulic device is arranged in the tundish. When the height of the stopper is adjusted by the hydraulic device, the flow of liquid steel from the tundish to the mold will change, as shown in Figure 3 The method comprises the following steps:

[0077] In step 301, historical data in the running process of the continuous casting billet liquid steel pouring system is collected. The historical data includes instantaneous liquid level historical values of liquid steel in the mold and stopper height historical set values of the stopper in the tundish at different historical moments. The instantaneous liquid level historical values include instantaneous liquid level historical values when the liquid level state of the liquid steel in the mold is normal and abnormal, respectively.

[0078] In step 302, based on the historical data after data preprocessing, a historical data training set, a historical data test set and a historical data validation set are obtained.

[0079] In step 303, based on the historical data training set, the historical data test set and the historical data validation set, a stopper height prediction model is trained.

[0080] In the above embodiment of the present application, first, historical data in the running process of the continuous casting billet liquid steel pouring system is collected. Based on the historical data after data preprocessing, a historical data training set, a historical data test set and a historical data validation set are obtained. Based on the historical data training set, the historical data test set and the historical data validation set, a stopper height prediction model is trained.

[0081] Specifically, the historical data can be obtained by on-site sampling, for example, the liquid level fluctuation and related process data can be extracted in the crystallizer of the continuous casting liquid pouring system of the steel plant, including the liquid level fluctuation data (instantaneous liquid level historical value) and stopper data (stopper height historical set value) at different historical moments, then the liquid level fluctuation data and the stopper data are respectively preprocessed, that is, the interference data in the original historical data is removed based on metallurgical mechanism and industrial mechanism, for example, according to the actual production conditions on site, abnormal fluctuation data caused by changes in process conditions such as package replacement, speed reduction and speed increase are removed, then the liquid level fluctuation data and the stopper data are respectively subjected to difference processing, and the data after difference processing is divided into a training set for training, a verification set for verification and a test set for testing, the BI-LSTM model is used, the liquid level fluctuation data after difference processing is taken as input, the preset liquid level prediction model is trained to predict the liquid level fluctuation, and the predicted liquid level fluctuation data is taken as input of the preset stopper prediction model, so as to predict the compensation position of the stopper in advance to prevent abnormal fluctuation. To this end, the trained preset liquid level prediction model and the preset stopper prediction model form a cycle prediction model to predict the liquid level fluctuation data under different steel grades and process conditions. In particular, the cycle prediction model can also be adjusted and optimized based on the prediction result, that is, according to the actual detection situation, it is determined whether the cycle prediction model needs to be updated. If the actual detection effect is not good, the data can be collected on site, and the preset liquid level prediction model and the preset stopper prediction model can be iteratively trained to improve the prediction accuracy of the preset liquid level prediction model and the preset stopper prediction model. By establishing a deep learning model, the instantaneous abnormal fluctuation of the liquid level in the crystallizer can be accurately predicted, and based on the liquid level prediction result, the stopper control can be compensated in advance to prevent the occurrence of instantaneous abnormal fluctuation of the liquid level in the crystallizer. In the actual prediction process of the cycle prediction model, if the actual detection effect is not good, the data can be collected on site, and the model can be iteratively trained to improve the prediction accuracy of the model.

[0082] In step 304, the instantaneous liquid level value of the molten steel in the crystallizer is collected in real time, and the collected instantaneous liquid level value is filtered and interpolated to obtain a to-be-predicted instantaneous liquid level value.

[0083] Then, the instantaneous liquid level value of the molten steel in the crystallizer is collected in real time. It is not clear that the one-to-one correspondence between the liquid level fluctuation and the stopper position is not clear, so the liquid level fluctuation data and the stopper position data are respectively subjected to difference processing to explore the relationship between them, that is, the collected instantaneous liquid level value is filtered and interpolated to obtain a to-be-predicted instantaneous liquid level value, which prepares for the prediction of the subsequent instantaneous liquid level prediction value.

[0084] At step 305, every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted based on a preset liquid level prediction model and the instantaneous liquid level values collected in a historical preset collection period, to obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0085] At step 306, when the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the instantaneous liquid level prediction value is subjected to fuzzy control based on a fuzzy proportional-integral-derivative controller.

[0086] Then, every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted based on a preset liquid level prediction model and the instantaneous liquid level values collected in a historical preset collection period, to obtain an instantaneous liquid level prediction value, and when the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the instantaneous liquid level prediction value is subjected to fuzzy control based on a fuzzy proportional-integral-derivative controller. In actual control of the instantaneous liquid level of the crystallizer, a fuzzy PID controller (Fuzzy PID Controller, PID, Proportional-Integral-Derivative) can also be used for control, so as to ensure that the compensation is in place and abnormal fluctuations in the liquid level are avoided. Specifically, in order to accurately realize compensation of the liquid level fluctuations, a fuzzy rule can also be designed to optimize the aforementioned fuzzy PID controller, the input of the fuzzy PID controller is determined as the difference e between the actual value and the set value of the liquid level change of the crystallizer and the change rate ec thereof, then the output domain parameters of the fuzzy parameter adjuster in the fuzzy PID controller can be determined as Δkp=[-0.7, 0.7], Δki=[-1.1, 1.1] and Δkd=[-0.21, 0.21], respectively, and the fuzzy subsets of the input and output of the fuzzy PID controller are defined as: negative huge (NH), negative medium (NM), negative small (NS), zero (O), positive small (PS), positive medium (PM) and positive huge (PH). Then, the fuzzy PID controller is subjected to fuzzy reasoning and defuzzification, in particular, the Mamdani reasoning method can be used to obtain the output of the fuzzy PID controller, and finally, the defuzzification is performed based on the center of gravity method to obtain the parameters. The architecture of the fuzzy PID controller is shown in Figure 4

[0087] ​Step 307, based on the stopper height prediction model and the fuzzy controlled instantaneous liquid level prediction value, predict the stopper height prediction set value that makes the abnormal liquid level state return to normal, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0088] Step 308, call the hydraulic device to adjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer that changes the molten steel flow is re-predicted based on the preset liquid level prediction model, and the molten steel liquid level state at the future preset time interval in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

[0089] Step 309, real-time monitoring of the height of the stopper in the tundish, when the monitoring of the height of the stopper in the tundish after the adjustment is not in the predicted stopper height prediction set value, readjust the height of the stopper in the tundish through the hydraulic device and the stopper height prediction set value.

[0090] Then, based on the stopper height prediction model and the fuzzy controlled instantaneous liquid level prediction value, predict the stopper height prediction set value that makes the abnormal liquid level state return to normal, call the hydraulic device to adjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer that changes the molten steel flow is re-predicted based on the preset liquid level prediction model, and the molten steel liquid level state at the future preset time interval in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value. Real-time monitoring of the height of the stopper in the tundish, when the monitoring of the height of the stopper in the tundish after the adjustment is not in the predicted stopper height prediction set value, readjust the height of the stopper in the tundish through the hydraulic device and the stopper height prediction set value. For this purpose, through real-time monitoring and fault diagnosis of the working state of the hydraulic device, the normal operation and accurate control of the hydraulic device can be ensured, and at the same time, according to the performance characteristics and process requirements of the hydraulic device, the adjustment range and precision of the hydraulic device can be reasonably set to meet the production requirements.

[0091] By applying the technical solution of the embodiment, historical data in the running process of the continuous casting billet liquid steel pouring system is collected, and based on the historical data after data preprocessing, a historical data training set, a historical data test set and a historical data validation set are obtained. Based on the historical data training set, the historical data test set and the historical data validation set, a stopper height prediction model is trained. The instantaneous liquid level value of the molten steel in the crystallizer is collected in real time, and the collected instantaneous liquid level value is filtered and interpolated to obtain a to-be-predicted instantaneous liquid level value. Based on the preset liquid level prediction model and the to-be-predicted instantaneous liquid level value collected in the preset collection period, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted to obtain an instantaneous liquid level prediction value. Whenever the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the instantaneous liquid level prediction value is controlled based on a fuzzy proportional-integral-derivative controller. Based on the stopper height prediction model and the fuzzy-controlled instantaneous liquid level prediction value, a stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted, and the hydraulic device is called to adjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow based on the preset liquid level prediction model is re-predicted, and the molten steel liquid level state in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value. The height of the stopper in the tundish is monitored in real time, and when it is monitored that the height value of the re-adjusted stopper in the tundish is not at the predicted stopper height prediction set value, the height of the stopper in the tundish is adjusted again by the hydraulic device and the stopper height prediction set value. The combination of the two prediction models can improve the stability of the liquid level fluctuation in the production process.

[0092] Further, as a refinement and expansion of the above embodiment, in order to fully describe the specific implementation process of the embodiment, another crystallizer instantaneous liquid level control method is provided, which is applied to a continuous casting billet liquid steel pouring system. The continuous casting billet liquid steel pouring system includes a ladle, a tundish and a crystallizer in sequence along the liquid steel pouring direction. The tundish is provided with a stopper whose height can be adjusted by a hydraulic device. When the height of the stopper is adjusted by the hydraulic device, the flow of the molten steel from the tundish to the crystallizer will change, as shown in Figure 5 The method comprises:

[0093] Step 401, the instantaneous liquid level value of the molten steel in the crystallizer is collected in real time.

[0094] In the above embodiment of the application, the instantaneous liquid level value of the molten steel in the crystallizer is collected in real time. For example, the on-site continuous casting production data can be collected in real time, including liquid level fluctuation data, stopper position data and different position argon blowing amount, etc. The composition of the steel grade of the continuous casting billet is shown in Table 1, and the related parameters of the continuous casting machine are shown in Table 2.

[0095] Table I

[0096]

[0097] Table II

[0098] Name Parameter Continuous casting machine model Vertical bending type Mold length (mm) 900 Mold oscillation frequency (times / min) 25-400 Mold oscillation amplitude (mm) 2-10 Casting strand width (mm) 1065-1522 Casting strand thickness (mm) 230 Drawing speed (m / min) 0.9-1.3

[0099] At step 402, every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted based on a preset liquid level prediction model and the instantaneous liquid level values collected in a preset historical collection period, to obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0100] Then, every preset time interval, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted based on a preset liquid level prediction model and the instantaneous liquid level values collected in a preset historical collection period, to obtain an instantaneous liquid level prediction value, for preparing for subsequent abnormal liquid level advance compensation adjustment.

[0101] At step 403, historical normal instantaneous liquid level values at multiple historical time points during the operation of the continuous casting billet molten steel pouring system are obtained, and a liquid level normal fluctuation range is determined based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values include multiple different values.

[0102] Then, historical normal instantaneous liquid level values at multiple historical time points during the operation of the continuous casting billet molten steel pouring system are obtained, and a liquid level normal fluctuation range is determined based on the historical normal instantaneous liquid level values, for example, the field liquid level fluctuation set value is 820 mm, and the liquid level normal fluctuation range can be set to 815 mm-825 mm.

[0103] At step 404, the liquid level state of the molten steel in the crystallizer at a future preset time interval is determined based on the instantaneous liquid level prediction value and the preset liquid level normal fluctuation range, wherein the liquid level state includes abnormal and normal.

[0104] At step 405, when the instantaneous liquid level prediction value is outside the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be abnormal, and when the instantaneous liquid level prediction value is within the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be normal.

[0105] Step 406, whenever the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0106] Step 407, the hydraulic device is called to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the molten steel liquid level state in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

[0107] Then, based on the instantaneous liquid level prediction value and the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined. When the instantaneous liquid level prediction value is outside the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be abnormal. When the instantaneous liquid level prediction value is within the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined to be normal. Whenever the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction set value that makes the abnormal liquid level state return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value. Finally, the hydraulic device is called to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the molten steel liquid level state in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

[0108] Specifically, the output of the preset liquid level prediction model is used as the input of the preset stopper height prediction model by combining the preset liquid level prediction model with the preset stopper height prediction model. Meanwhile, the liquid surface fluctuation range within 815mm-825mm is set as normal fluctuation, and the rest is abnormal fluctuation. When the liquid surface fluctuation prediction result is abnormal fluctuation, it is input into the stopper position prediction model to realize the advance compensation of the liquid surface fluctuation and prevent the generation of abnormal fluctuation.

[0109] Further, as Figure 1 the specific implementation of the method, the embodiment of the application provides a crystallizer instantaneous liquid surface control device, as shown in the figure, the device comprises: Figure 6

[0110] The on-site liquid level acquisition module 501 is used for acquiring the instantaneous liquid level value of the molten steel in the crystallizer in real time.

[0111] ​The future liquid level prediction module 502 is configured to, at every preset time interval, predict an instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and an instantaneous liquid level value collected in a preset historical collection period, and obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0112] The abnormal liquid level stopper height prediction module 503 is configured to, when determining that the liquid level state of the molten steel in the crystallizer at the future preset time interval is abnormal based on the predicted instantaneous liquid level prediction value, predict a stopper height prediction set value that makes the abnormal liquid level state return to normal based on a stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0113] The stopper height adjustment module 504 is configured to call the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer in which the molten steel flow changes is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer at the future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

[0114] Optionally, the future liquid level prediction module 502 is further configured to:

[0115] At every preset time interval, predict an instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and an instantaneous liquid level value collected in a historical preset collection period, and obtain an instantaneous liquid level prediction value.

[0116] Optionally, the abnormal liquid level stopper height prediction module 503 is further configured to:

[0117] Based on the stopper height prediction model and the fuzzy-controlled instantaneous liquid level prediction value, predict a stopper height prediction set value that makes the abnormal liquid level state return to normal.

[0118] Further, the embodiment of the present application provides a crystallizer instantaneous liquid level control device, as shown in the accompanying drawings, Figure 7 The device comprises:

[0119] The on-site liquid level collection module 601 is configured to collect an instantaneous liquid level value of the molten steel in the crystallizer in real time;

[0120] The future liquid level prediction module 602 is configured to, at every preset time interval, predict an instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and an instantaneous liquid level value collected in a preset historical collection period, and obtain an instantaneous liquid level prediction value, wherein the preset liquid level prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0121] The abnormal liquid level stopper height prediction module 603 is configured to, when it is determined that the liquid level state of the molten steel in the crystallizer at the future preset time interval is abnormal based on the predicted instantaneous liquid level prediction value, predict a stopper height prediction set value that makes the abnormal liquid level state return to normal based on a stopper height prediction model and the instantaneous liquid level prediction value, wherein the stopper height prediction model is established based on a BI-LSTM (Bidirectional Long Short-Term Memory) network structure.

[0122] The stopper height adjustment module 604 is configured to call the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer in which the molten steel flow changes is re-predicted based on the preset liquid level prediction model, and it is determined that the liquid level state of the molten steel in the crystallizer at the future preset time interval is normal based on the re-predicted instantaneous liquid level prediction value.

[0123] The stopper height prediction model training module 605 is configured to: collect historical data in a running process of the continuous casting billet molten steel pouring system, wherein the historical data includes instantaneous liquid level historical values of the molten steel in the crystallizer and stopper height historical set values of the stopper in the tundish at different historical time points, and the instantaneous liquid level historical values include instantaneous liquid level historical values when the liquid level state of the molten steel in the crystallizer is normal and abnormal, respectively; obtain a historical data training set, a historical data test set and a historical data validation set based on the historical data after data preprocessing; and train the stopper height prediction model based on the historical data training set, the historical data test set and the historical data validation set.

[0124] The data preprocessing module 606 is configured to perform filtering processing and interpolation processing on the collected instantaneous liquid level value to obtain a to-be-predicted instantaneous liquid level value.

[0125] The fuzzy control module 607 is configured to perform fuzzy control on the instantaneous liquid level prediction value based on a fuzzy proportional-integral-derivative controller.

[0126] The abnormal liquid level determination module 608 is configured to determine a liquid level state of the molten steel in the crystallizer at a future preset time interval based on the instantaneous liquid level prediction value and a preset liquid level normal fluctuation range, wherein the liquid level state comprises an abnormal state and a normal state; when the instantaneous liquid level prediction value is outside the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined as abnormal; and when the instantaneous liquid level prediction value is within the preset liquid level normal fluctuation range, the liquid level state of the molten steel in the crystallizer is determined as normal.

[0127] The stopper height monitoring module 609 is configured to obtain historical normal instantaneous liquid level values at a plurality of historical time points during operation of the continuous casting billet molten steel pouring system, and determine the liquid level normal fluctuation range based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values comprise a plurality of different values.

[0128] Optionally, the future liquid level prediction module 602 is further configured to:

[0129] At every preset time interval, the future liquid level prediction module 602 is configured to predict the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and a to-be-predicted instantaneous liquid level value collected in a preset collection period, to obtain an instantaneous liquid level prediction value.

[0130] Optionally, the abnormal liquid level stopper height prediction module 603 is further configured to:

[0131] The abnormal liquid level stopper height prediction module 603 is configured to predict a stopper height prediction set value that makes the abnormal liquid level state return to normal based on a stopper height prediction model and the fuzzy-controlled instantaneous liquid level prediction value.

[0132] Optionally, the abnormal liquid level determination module 608 is further configured to:

[0133] The abnormal liquid level determination module 608 is configured to obtain historical normal instantaneous liquid level values at a plurality of historical time points during operation of the continuous casting billet molten steel pouring system, and determine the liquid level normal fluctuation range based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values comprise a plurality of different values.

[0134] It should be noted that other corresponding descriptions of the functions of the crystallizer instantaneous liquid level control device provided in the embodiments of the present application can be referred to the corresponding descriptions in the Figure 1 、 Figure 3 and Figure 5 methods, which will not be described herein again.

[0135] Based on the above-mentioned methods as shown in Figure 1 、 Figure 3 and Figure 5 , correspondingly, the embodiments of the present application further provide a storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the above-mentioned methods as shown in Figure 1 ,Figure 3 and Figure 5 the crystallizer instantaneous liquid level control method shown in

[0136] Based on such understanding, the technical scheme of the present application can be embodied in the form of a software product, which can be stored in a nonvolatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method described in various implementation scenarios of the present application.

[0137] Based on the method shown in Figure 1 , Figure 3 and Figure 5 and Figure 6 and Figure 7 In order to achieve the above-mentioned purpose, the embodiments of the present application further provide a computer device, which can be a personal computer, a server, a network device, etc., and the computer device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to realize the crystallizer instantaneous liquid level control method shown in Figure 1 , Figure 3 and Figure 5 .

[0138] Optionally, the computer device can further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface can include a display, an input unit such as a keyboard, etc. The optional user interface can further include a USB interface, a card reader interface, etc. The network interface can optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.

[0139] Those skilled in the art can understand that the structure of the computer device provided by the embodiments does not constitute a limitation on the computer device, and can include more or fewer components, or combine certain components, or different component arrangements.

[0140] The storage medium can further include an operating system and a network communication module. The operating system is a program for managing and saving computer device hardware and software resources, supporting information processing programs and the running of other software and / or programs. The network communication module is used to realize the communication between the components in the storage medium, and the communication with other hardware and software in the entity device.

[0141] Through the description of the above embodiments, those skilled in the art can clearly understand that the application can be realized by means of software and necessary general hardware platforms, and the real-time acquisition of the instantaneous liquid level value of the molten steel can also be realized by hardware; based on the preset liquid level prediction model and the instantaneous liquid level value, the instantaneous liquid level prediction value of the molten steel is predicted; whenever the liquid level state of the molten steel in the crystallizer is determined to be abnormal based on the predicted instantaneous liquid level prediction value, the stopper height prediction model and the instantaneous liquid level prediction value are used to predict the stopper height prediction set value, so that the hydraulic device adjusts the height of the stopper in the tundish based on the predicted stopper height prediction set value until the crystallizer with changed molten steel flow is re-predicted based on the preset liquid level prediction model, and the liquid level state of the molten steel in the crystallizer is determined to be normal based on the re-predicted instantaneous liquid level prediction value. By combining the two prediction models to obtain a cyclic prediction model, the liquid level fluctuation stability in the production process can be improved.

[0142] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the drawings are not necessarily required for implementing the application. Those skilled in the art can understand that the modules in the device in the implementation scenario can be distributed in the device in the implementation scenario according to the description of the implementation scenario, or can be changed and located in one or more devices different from the implementation scenario. The modules of the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0143] The above application numbers are only for description, and do not represent the advantages and disadvantages of the implementation scenario. The above disclosure is only a few specific implementation scenarios of the application, but the application is not limited thereto, and any changes that can be thought of by those skilled in the art should fall within the protection scope of the application.

Claims

1. A method for instantaneous liquid level control in a crystallizer, characterized in that, An application is made to a continuous casting billet molten steel pouring system, wherein the continuous casting billet molten steel pouring system includes, in sequence along the molten steel pouring direction, a ladle, a tundish, and a crystallizer. The tundish is equipped with a stopper rod whose height can be adjusted by a hydraulic device. When the height of the stopper rod is adjusted by the hydraulic device, the flow rate of molten steel from the tundish to the crystallizer will change. The method includes: Real-time acquisition of the instantaneous liquid level of molten steel in the crystallizer; Every preset time interval, based on the preset liquid level prediction model and the instantaneous liquid level values ​​collected within the preset historical collection period, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted to obtain the instantaneous liquid level prediction value. The preset liquid level prediction model is established based on the BI-LSTM (bidirectional long short-term memory) network structure. Whenever the liquid level of the molten steel in the crystallizer is determined to be abnormal at a future preset time interval based on the predicted instantaneous liquid level, a stopper height prediction set value that would cause the abnormal liquid level to return to normal is predicted based on the stopper height prediction model and the instantaneous liquid level prediction value. The stopper height prediction model is established based on a BI-LSTM (bidirectional long short-term memory) network structure. The hydraulic device is invoked to readjust the height of the stopper rod in the tundish based on the predicted stopper rod height prediction set value, until the crystallizer that changes the molten steel flow rate is re-predicted based on the preset liquid level prediction model, and the molten steel level state in the crystallizer at a future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

2. The method according to claim 1, characterized in that, Before predicting the stopper rod height prediction setpoint that would cause the abnormal liquid level state to return to normal, based on the stopper rod height prediction model and the instantaneous liquid level prediction value, the method further includes: Historical data is collected during the operation of the continuous casting billet molten steel pouring system. The historical data includes the historical values ​​of the instantaneous liquid level of the molten steel in the crystallizer and the historical set value of the stopper height of the stopper in the tundish at different historical times. The historical values ​​of the instantaneous liquid level include the historical values ​​of the instantaneous liquid level when the liquid level of the molten steel in the crystallizer is in a normal and abnormal state, respectively. Based on the preprocessed historical data, obtain the historical data training set, historical data test set, and historical data validation set; The stopper height prediction model is trained based on the historical data training set, the historical data test set, and the historical data validation set.

3. The method according to claim 1, characterized in that, After acquiring the instantaneous liquid level value of the molten steel in the real-time crystallizer, the method further includes: The collected instantaneous liquid level values ​​are filtered and interpolated to obtain the instantaneous liquid level value to be predicted. Accordingly, at each preset time interval, based on a preset liquid level prediction model and the instantaneous liquid level values ​​collected within a preset historical collection period, the instantaneous liquid level value of the molten steel in the crystallizer is predicted at a future preset time interval to obtain an instantaneous liquid level prediction value, including: At preset time intervals, based on the preset liquid level prediction model and the instantaneous liquid level values ​​collected during the preset historical collection period, the instantaneous liquid level value of the molten steel in the crystallizer at a future preset time interval is predicted, and the instantaneous liquid level prediction value is obtained.

4. The method according to claim 1, characterized in that, Before predicting the stopper rod height prediction setpoint that would cause the abnormal liquid level state to return to normal, based on the stopper rod height prediction model and the instantaneous liquid level prediction value, the method further includes: Fuzzy control of instantaneous liquid level prediction based on fuzzy proportional-integral-derivative controller; Accordingly, the prediction of the stopper height prediction setpoint that causes the abnormal liquid level state to return to normal, based on the stopper height prediction model and the instantaneous liquid level prediction value, includes: Based on the stopper height prediction model and the instantaneous liquid level prediction value after fuzzy control, the stopper height prediction setpoint that causes the abnormal liquid level state to return to normal is predicted.

5. The method according to claim 1, characterized in that, Before determining that the liquid level state of the molten steel in the crystallizer is abnormal at a future preset time interval based on the predicted instantaneous liquid level value, the method further includes: Based on the instantaneous liquid level prediction value and the preset normal fluctuation range of liquid level, the liquid level status of the molten steel in the crystallizer at a future preset time interval is determined, wherein the liquid level status includes abnormal and normal. When the instantaneous liquid level prediction value is outside the preset normal fluctuation range of liquid level, the liquid level state of the molten steel in the crystallizer is determined to be abnormal; when the instantaneous liquid level prediction value is within the preset normal fluctuation range of liquid level, the liquid level state of the molten steel in the crystallizer is determined to be normal.

6. The method according to claim 1, characterized in that, The method further includes: During the operation of the continuous casting billet molten steel pouring system, the historical normal instantaneous liquid level values ​​at multiple historical moments are obtained, and the normal fluctuation range of the liquid level is determined based on the historical normal instantaneous liquid level values, wherein the historical normal instantaneous liquid level values ​​include multiple different values.

7. The method according to claim 1, characterized in that, After the method involves invoking the hydraulic device to readjust the height of the stopper rod in the tundish based on the predicted stopper rod height setpoint, the method further includes: The height of the stopper rod in the tundish is monitored in real time. When the height of the readjusted stopper rod in the tundish is not at the predicted stopper rod height setting value, the height of the stopper rod in the tundish is readjusted through the hydraulic device and the stopper rod height prediction setting value.

8. A device for instantaneous liquid level control in a crystallizer, characterized in that, An apparatus for use in a continuous casting billet molten steel pouring system, wherein the continuous casting billet molten steel pouring system comprises, in sequence along the pouring direction, a ladle, a tundish, and a crystallizer. The tundish is equipped with a stopper rod whose height can be adjusted by a hydraulic device. When the height of the stopper rod is adjusted by the hydraulic device, the flow rate of molten steel from the tundish to the crystallizer will change. The apparatus includes: The on-site liquid level acquisition module is used to acquire the instantaneous liquid level value of molten steel in the crystallizer in real time; The future liquid level prediction module is used to predict the instantaneous liquid level of the molten steel in the crystallizer at a future preset time interval based on a preset liquid level prediction model and the instantaneous liquid level values ​​collected during a preset historical collection period, so as to obtain the instantaneous liquid level prediction value. The preset liquid level prediction model is based on a BI-LSTM (bidirectional long short-term memory) network structure. The abnormal liquid level stopper height prediction module is used to predict a stopper height prediction set value that will cause the abnormal liquid level state to return to normal whenever the liquid level state of the molten steel in the crystallizer is determined to be abnormal at a future preset time interval based on the predicted instantaneous liquid level prediction value. The stopper height prediction model is based on a BI-LSTM (bidirectional long short-term memory) network structure. The stopper height adjustment module is used to call the hydraulic device to readjust the height of the stopper in the tundish based on the predicted stopper height setting value, until the crystallizer that changes the molten steel flow rate is re-predicted based on the preset liquid level prediction model, and the molten steel level status in the crystallizer at a future preset time interval is determined to be normal based on the re-predicted instantaneous liquid level prediction value.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for instantaneous liquid level control of the crystallizer as described in any one of claims 1 to 7.

10. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for instantaneous liquid level control of the crystallizer as described in any one of claims 1 to 7.

Images