Method for controlling cooling flow rate in wire rod rolling process using artificial intelligence

Abstract

The present invention provides a method for controlling the cooling flow rate of a wire rod rolling process, comprising: (a) a step of selecting process data regarding the temperature of the material before finishing rolling in a hot rolling process; (b) a step of collecting past process performance data regarding the process data and training an artificial intelligence model; (c) a step of calculating a cooling water injection flow rate value using the trained artificial intelligence model when extracting the material; and (d) a step of spraying cooling water onto the material according to the calculated cooling water injection flow rate value.

Description

Technology Field

[0001] The present invention relates to water cooling control in a wire rod hot rolling process using artificial intelligence, and specifically to a method of controlling the cooling flow rate of a water cooling facility to converge the material temperature to a predetermined target temperature for each process section. Background Technology

[0002] In the wire rod hot rolling process, billets used as materials are generally heated to approximately 950–1200°C in a reheating furnace, then undergo rough rolling, intermediate rolling, finishing rolling, and final rolling processes, and are produced by coiling at approximately 850–950°C and undergoing an air cooling process. During this process, the material temperature for each section is controlled by applying a water cooling process immediately before rolling and coiling following the rough and intermediate rolling stages, in accordance with product design standards.

[0003] Figure 1 is a flowchart of the cooling flow rate control for a wire rod rolling process according to a conventional method. According to the conventional method, the flow rate of cooling water sprayed onto the material must be appropriately controlled to converge the material to each target temperature. Typically, approximate water injection standards are set based on the production steel grade, size, and furnace extraction temperature, and applied during material extraction and rolling (Figure 1b). Subsequently, the operator checks the measured material temperature for each section and adjusts the initially set flow rate so that the material converges to the target temperature. After convergence, the flow rate control is automated by applying PID control to the material being continuously rolled (Figure 1a).

[0004] The time required from the first material extraction to convergence to the target temperature may vary depending on the equipment operator's control method; while convergence may occur as early as the first material, if it does not, the time required to reach the target temperature and the number of rolled materials may increase. This issue may result in failure to meet the required quality standards for the product and can cause quality variations between the entire material and the produced products.

[0005] In the case of Korean published patent No. 10-2001-0010808, the goal was to shorten the time required for feedback by utilizing previous water cooling performance and improving equipment controllability as a method to shorten the time required to reach the target flow rate and control stabilization time, but there has not yet been a case of approaching it by predicting the required cooling flow rate by utilizing process conditions obtained in real time. Prior art literature

[0006] Korean Published Patent No. 10-2001-0010808 The problem to be solved

[0007] The present invention aims to solve such conventional problems by utilizing a prediction model based on real-time process data, rather than a method of setting the initial flow rate using previous and past water cooling performance, thereby minimizing the convergence time to a target range for material cooling. However, this objective is exemplary and does not limit the scope of the present invention. means of solving the problem

[0008] According to one aspect of the present invention, a method for controlling the cooling flow rate of a wire rod rolling process is provided.

[0009] According to one embodiment, the cooling flow rate control method may include, when cooling a material prior to finishing rolling in a hot rolling process, (a) a step of selecting process data regarding the material temperature; (b) a step of collecting past process performance data regarding the process data and training an artificial intelligence model; (c) a step of calculating a cooling water injection flow rate value using the trained artificial intelligence model when extracting the material; and (d) a step of injecting cooling water onto the material according to the calculated cooling water injection flow rate value.

[0010] According to one embodiment, the process data regarding the material temperature may include material specification information, heating furnace extraction conditions, descaler operation conditions, rough rolling cooling conditions, and water cooling process conditions.

[0011] According to one embodiment, the artificial intelligence model may be any one selected from a linear model, an XGBoost model, a LightGBM model, an MLP model, and a combination thereof.

[0012] According to one embodiment, step (c) may include the step of transmitting furnace process data and post-process data to an artificial intelligence model when the material is extracted from the furnace.

[0013] According to one embodiment, the furnace process data includes material specification information and furnace extraction conditions, and the post-process data may include descaler operation conditions, rough rolling cooling conditions, and water cooling process conditions.

[0014] According to one embodiment, step (d) may include a step in which a calculated cooling water injection flow rate value is input to an equipment control unit. Effects of the invention

[0015] According to an embodiment of the present invention as described above, in water cooling control for a material at the beginning of a rolling operation, the resources and time required for convergence to a target temperature can be reduced.

[0016] In addition, by training an artificial intelligence model with data on materials produced in the past, the predictive performance for new data can be improved. This enables the material temperature to converge to a target range in a short period of time, and allows for the resolution of issues such as quality non-uniformity caused by temperature variations between or within materials, as well as reduced rollability.

[0017] Furthermore, by partially replacing the role of equipment operators, convenience in operational tasks can be enhanced, and by extending the application of this technology to similar process control areas, a foundation for the transition to a smart factory can be laid.

[0018] Of course, the scope of the present invention is not limited by these effects. Brief explanation of the drawing

[0019] Figure 1 is a flowchart of the cooling flow rate control for a wire rod rolling process according to a conventional method. Figure 2 is a flowchart of the cooling flow rate control for a wire rod rolling process according to an embodiment of the present invention. Figure 3 is a data flow diagram for controlling the cooling flow rate of a wire rod rolling process according to an embodiment of the present invention. Figure 4 is the result of a prediction performance test using an artificial intelligence model according to an embodiment of the present invention. Figure 5 shows the results of performance improvement using an artificial intelligence model according to an embodiment of the present invention. Specific details for implementing the invention

[0020] Hereinafter, various embodiments of the present invention will be described in detail with reference to the attached drawings. The embodiments of the present invention are provided to more fully explain the present invention to those skilled in the art, and the following embodiments may be modified in various different forms, and the scope of the present invention is not limited to the following embodiments. Rather, these embodiments are provided to make the present disclosure more faithful and complete and to fully convey the spirit of the present invention to those skilled in the art. In addition, the thickness or size of each layer in the drawings is exaggerated for convenience and clarity of explanation.

[0021] The present invention is a technology for cooling a material before finishing rolling in a hot rolling process, which learns existing water cooling performance using artificial intelligence algorithms such as artificial neural networks and tree-based ensembles, and controls the flow rate based on data-based decision-making rather than the subjective judgment of the equipment operator.

[0022] To achieve the objective of the present invention, the first step is to collect and learn process data. First, process data that may affect cooling and material temperature from the start to the end of rolling is selected, and all past process performance data is collected accordingly and used to train an artificial intelligence model.

[0023] The above process data includes multiple pieces of information related to material temperature, such as material specification information, furnace extraction conditions, descaler operation conditions, rough rolling cooling conditions, and water cooling process conditions. The above process data may include any information regarding material temperature, but may be excluded if it is determined to have low relevance to material temperature. For example, information such as rolling roll speed and roll usage may be excluded.

[0024] Finally, the parameters of the model that has been trained and optimized are stored on a separate server for utilizing the artificial intelligence model and are updated periodically. Using the trained artificial intelligence model, the cooling water injection flow rate value can be calculated from the process data. Accordingly, the cooling flow rate control device according to the embodiment of the present invention may be referred to as an AI server.

[0025] The cooling flow control device transmits the calculated cooling water injection flow rate value to the equipment control unit. The equipment can spray cooling water onto the material according to the received cooling water injection flow rate value information.

[0026] Hereinafter, after briefly describing the artificial intelligence model, a cooling flow rate control method according to one embodiment of the present invention will be described in more detail.

[0027] Artificial intelligence (AI) is a field of computer science and information technology that studies methods to enable computers to perform thinking, learning, and self-development capable of human intelligence, and refers to the ability of computers to mimic intelligent human behavior.

[0028] Machine learning is a field of artificial intelligence that enables computers to learn without explicit programming.

[0029] Specifically, machine learning is a technology that studies and builds systems and algorithms capable of learning, making predictions, and improving their own performance based on empirical data. Rather than executing strictly defined static program commands, machine learning algorithms adopt an approach of constructing specific models to derive predictions or decisions based on input data.

[0030] Meanwhile, according to one embodiment of the present invention, the artificial intelligence model used in the cooling flow rate control method may include at least one of a linear model, an XGBoost algorithm model, a LightGBM algorithm model, and an MLP algorithm model. In addition, not necessarily limited thereto, various forms of artificial intelligence, deep learning, or machine learning models, such as ANN models, CNN models, RNN models, and SLP models, may all be applied.

[0031] The XGBoost and LightGBM algorithm models are boosting models. Boosting models are a type of ensemble technique that combines multiple weak decision trees; they are algorithmic models that build strong prediction models by weighting the learning errors of weak prediction models and sequentially applying them to the next training model.

[0032] In particular, the algorithm model used in the embodiment of the present invention may be a combination model of a linear model, an XGBoost algorithm model, a LightGBM algorithm model, and an MLP algorithm model, i.e., an ensemble model.

[0033] Ensemble models solve problems by training and combining multiple algorithm models, rather than training and using only a single algorithm model. In particular, ensemble learning is effective when the desired performance cannot be achieved with a single algorithm model, and it can improve generalization performance by combining multiple algorithm models trained individually.

[0034] FIG. 2 is a flowchart of the cooling flow rate control of a wire rod rolling process according to an embodiment of the present invention, and FIG. 3 is a data flowchart according to the same.

[0035] The process of controlling the cooling flow rate for each step illustrated in Figures 2 and 3 is explained in detail.

[0036] First, when the first material is extracted during product production, furnace process data and subsequent process data are simultaneously transmitted as input values ​​to an artificial intelligence model. The furnace process data may include material specification information and furnace extraction conditions. The subsequent process data is a control value set in equipment located downstream of the furnace and may include descaler operation conditions, coarse / intermediate rolling cooling conditions, finishing / precision rolling cooling conditions, and water cooling process conditions.

[0037] Next, the artificial intelligence model produces outputs, such as predicted values ​​like the cooling water injection flow rate, based on learned parameters. These outputs are directly input into the equipment control unit, and the water cooling system injects the cooling flow rate according to the input control values.

[0038] Figure 4 is a graph showing the performance results of various artificial intelligence models that can be applied in the hot rolling process.

[0039] As shown in Figure 4, when the Linear model, XGBoost model, LightGBM model, and MLP model are selected among artificial intelligence models and the NTM inlet temperature, RSM inlet temperature, and winding temperature of the material are compared with the predicted values, it can be confirmed that the material temperature generally converges to the target range.

[0040] Figure 5 is a graph showing the results of performance improvement compared to the existing model by selecting the LightGBM model among artificial intelligence models.

[0041] As shown in Figure 5, it was confirmed that the Spec-in Ratio for each section, that is, the probability of falling within the specifications, was 80 to 98 percent, which is significantly improved compared to the conventional 67 to 90 percent.

[0042] According to an embodiment of the present invention as described above, in water cooling control for a material at the beginning of a rolling operation, the resources and time required for convergence to a target temperature can be reduced.

[0043] In addition, by training an artificial intelligence model with data on materials produced in the past, the predictive performance for new data can be improved. This enables the material temperature to converge to a target range in a short period of time, and allows for the resolution of issues such as quality non-uniformity caused by temperature variations between or within materials, as well as reduced rollability.

[0044] Furthermore, by partially replacing the role of equipment operators, convenience in operational tasks can be enhanced, and by extending the application of this technology to similar process control areas, a foundation for the transition to a smart factory can be laid.

[0045] The present invention has been described with reference to the embodiments illustrated in the drawings, but this is merely illustrative, and those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible therefrom. Accordingly, the true technical scope of protection of the present invention should be determined by the technical spirit of the appended claims.

Claims

Claim 1 A method for controlling the cooling flow rate of a wire rod rolling process, comprising: (a) a step of selecting process data regarding the material temperature before finishing rolling in a hot rolling process; (b) a step of collecting past process performance data regarding the process data and training an artificial intelligence model; (c) a step of calculating a cooling water injection flow rate value using the trained artificial intelligence model when extracting the material; and (d) a step of spraying cooling water onto the material according to the calculated cooling water injection flow rate value; wherein step (c) includes a step of transmitting furnace process data and subsequent process data to an artificial intelligence model when the material is extracted from the furnace, wherein the furnace process data includes material specification information and furnace extraction conditions, and the subsequent process data includes descaler operation conditions, rough rolling cooling conditions, and water cooling process conditions, and step (c) calculates a cooling water injection flow rate value to converge the material temperature to a target range in the stage before finishing rolling using the artificial intelligence model. Claim 2 delete Claim 3 A method for controlling the cooling flow rate of a wire rod rolling process, wherein the artificial intelligence model is selected from a linear model, an XGBoost model, a LightGBM model, an MLP model, and a combination thereof. Claim 4 delete Claim 5 delete Claim 6 A method for controlling the cooling flow rate of a wire rod rolling process, wherein, in claim 1, step (d) includes the step of inputting the calculated cooling water injection flow rate value into an equipment control unit.

Images