Cold-rolled steel sheet material quality control system and cold-rolled steel sheet material quality prediction model generation method

Abstract

A system for controlling the material quality of a cold-rolled steel sheet according to an embodiment of the present invention includes a programmable logic controller that acquires process data of a cold-rolled steel sheet that has already been produced, a process controller that controls a manufacturing process of the cold-rolled steel sheet according to the process data of the programmable logic controller, a model generation unit that learns a learning data set in advance and generates a prediction model that predicts the material quality in the length direction of the cold-rolled steel sheet to be produced, and a production management system that transmits and receives information to and from the process controller, receives a predicted value of the prediction model generated by the model generation unit, and manages the production of the cold-rolled steel sheet to be produced. A method for generating a material quality prediction model for a cold-rolled steel sheet according to an embodiment of the present invention includes: a programmable logic controller that acquires process data of a cold-rolled steel sheet that has already been produced; a process controller that controls a manufacturing process of the cold-rolled steel sheet according to the process data of the programmable logic controller; a model generation unit that learns a learning data set in advance and generates a prediction model that predicts the material quality in the length direction of the cold-rolled steel sheet to be produced; and a production management system that transmits and receives information to and from the process controller, receives a predicted value of the prediction model generated by the model generation unit, and manages the production of the cold-rolled steel sheet to be produced. and generating a prediction model that predicts the longitudinal properties of the cold-rolled steel sheet to be produced by training the training dataset with a preset model, wherein the plurality of data included in the training dataset include hot rolling process data among process data provided from a programmable logic controller, material data and component data provided from a production management system, and data mapping measurement positions between cold rolling and annealing process data among process data from the programmable logic controller, and the process data provided by the programmable logic controller may be process data of the produced cold-rolled steel sheet acquired in advance by a sensor.

Description

[Technical Field]

[0001] The present invention relates to a system for controlling the material quality of a cold-rolled steel sheet and a method for generating a material quality prediction model for a cold-rolled steel sheet. [Background technology]

[0002] In recent years, countries around the world have been enacting strict policies regarding various environmental regulations, such as CO2 emission limits, and energy usage restrictions. Therefore, improving fuel efficiency and durability is a key issue that automakers must address. To achieve this, the use of thin, high-strength steel can simultaneously address various issues related to the environment, fuel efficiency, crashworthiness, and durability. High-strength steel continues to be developed to reduce vehicle weight. However, the development of high-strength steel requires increasing amounts of elements such as C, Mn, and Si. The higher the steel element content, the greater the material sensitivity (changes in material properties) to changes in process temperature. As a result, the advancement to giga-class high-strength steel leads to greater material variation within the coil. This variation in material properties leads to variations in springback within a single coil during client molding, resulting in persistent defects in molding quality. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Korean Patent Registration No. 10-0858902 Summary of the Invention [Problem to be solved by the invention]

[0004] According to one embodiment of the present invention, there are provided a system for controlling the quality of a cold-rolled steel sheet and a method for generating a model for predicting the quality of a cold-rolled steel sheet according to its entire length using artificial intelligence. [Means for solving the problem]

[0005] In order to solve the above-mentioned problems of the present invention, a material quality control system for cold-rolled steel sheets according to one embodiment of the present invention includes: a programmable logic controller that acquires process data of cold-rolled steel sheets that have already been produced; a process controller that controls a manufacturing process of the cold-rolled steel sheets in accordance with the process data of the programmable logic controller; a model generation unit that learns a learning dataset in advance and generates a prediction model that predicts the material quality in the longitudinal direction of the cold-rolled steel sheets to be produced; and a production management system that transmits and receives information with the process controller, receives predicted values ​​of the prediction model generated by the model generation unit, and manages the production of the cold-rolled steel sheets to be produced, wherein the learning dataset may include hot-rolling process data from the process data from the programmable logic controller, material quality data and component data from the production management system, and data mapping measurement positions between cold-rolling and annealing process data from the process data from the programmable logic controller.

[0006] A method for generating a material quality prediction model for a cold-rolled steel sheet according to one embodiment of the present invention includes the steps of: generating a training dataset by a model generation unit; and training a preset model on the training dataset to generate a prediction model for predicting the material quality in the longitudinal direction of a produced cold-rolled steel sheet. The plurality of data included in the training dataset include hot rolling process data among process data provided from a programmable logic controller, material data and component data provided from a production management system, and data mapping measurement positions between cold rolling and annealing process data among process data from the programmable logic controller. The process data provided by the programmable logic controller may be process data of the produced cold-rolled steel sheet acquired in advance by a sensor. [Effects of the Invention]

[0007] According to one embodiment of the present invention, it is possible to demonstrate precise and accurate prediction capabilities, which is effective in resolving factors that reduce production capacity, such as the need to modify molds due to plate material portions within coils during forming or changes in forming conditions. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic configuration diagram of a material quality control system for cold-rolled steel sheets according to an embodiment of the present invention. [Figure 2] FIG. 4 is a diagram showing process data detected in a material quality control system for cold-rolled steel sheets according to an embodiment of the present invention. [Figure 3] 1 is a schematic flowchart of a method for generating a material quality prediction model for a cold-rolled steel sheet according to an embodiment of the present invention. [Figure 4] FIG. 10 is a diagram showing learning data variables in the method for generating a material quality prediction model for a cold-rolled steel sheet according to an embodiment of the present invention. [Figure 5] FIG. 10 is a diagram showing weight settings in a method for generating a material quality prediction model for a cold-rolled steel sheet according to an embodiment of the present invention. [Figure 6] FIG. 10 is a diagram showing weight settings in a method for generating a material quality prediction model for a cold-rolled steel sheet according to an embodiment of the present invention. [Figure 7a] 3 is a flowchart relating to full length material quality control of the cold-rolled steel sheet material quality control system according to an embodiment of the present invention. [Figure 7b] FIG. 10 is a diagram showing an example in which an optimal annealing temperature is presented over the entire length in a material quality control system for cold-rolled steel sheets according to an embodiment of the present invention. [Figure 8] 1 is a diagram illustrating an exemplary computing environment in which a material property control system for cold-rolled steel sheets according to an embodiment of the present invention can be implemented. DETAILED DESCRIPTION OF THE INVENTION

[0009] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments will now be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention.

[0010] First, in a system for controlling the quality of cold-rolled steel sheets and a method for generating a quality prediction model for cold-rolled steel sheets according to an embodiment of the present invention, a cold-rolled steel sheet, for example, a cold-rolled steel sheet for automobiles, is produced by producing a steel slab containing, by weight, C: 0.001-0.4%, Si: 0.1-2.5%, Mn: 1-3%, Al: 0.1% or less, Cr: 1.0% or less, Ti: 0.2% or less, Nb: 0.1% or less, V: 0.2% or less, Mo: 0.5% or less, B: 30 ppm or less, with the balance being Fe and unavoidable impurities, and then heating the steel slab. This is to facilitate the subsequent hot rolling process and obtain the desired quality, and is preferably heated to a temperature range of 1100-1300°C. Hot rolling involves rough rolling followed by finish rolling, which then enters the coiling (CT) stage, where a cooling process is performed. The coiled hot-rolled steel sheet can be cold-rolled at room temperature with a predetermined reduction to produce a cold-rolled steel sheet. After cold rolling, the steel sheet is preferably subjected to continuous annealing. The continuous annealing may include a heating section (HS), a soaking section (SS), a slow cooling section (SCS), a rapid cooling section (RCS), a reheating section (RHS), an overaging section (OAS), and a final cooling section (FCS).

[0011] FIG. 1 is a schematic diagram of a system for controlling the quality of cold-rolled steel sheets according to one embodiment of the present invention.

[0012] Referring to FIG. 1 , a material quality control system for cold-rolled steel sheets according to an embodiment of the present invention may include a sensor 110, a programmable logic controller (PLC) 120, a process controller (PC) 130, a manufacturing execution system (MES) 140, and a model generation unit 150.

[0013] The sensors 110 are arranged in the above-mentioned cold-rolled steel sheet manufacturing process to detect process data, the programmable logic controller 120 acquires the process data from the sensors 110 and sends it to the process controller 130, the process controller 130 communicates with the production management system 140 to control the production management system 140, and controls the programmable logic controller 120 according to the acquired process data, thereby controlling the above-mentioned cold-rolled steel sheet manufacturing process. The programmable logic controller 120 can send the process data acquired from the sensors 110 to the model generation unit 150.

[0014] The model generation unit 150 generates a learning data set according to the process data and trains a pre-set model, thereby generating a prediction model that can predict the material properties of the cold-rolled steel sheet in the length direction (total length).

[0015] The model generating unit 150 may include first and second databases (DBs) 151 and 152 and a model generator 153.

[0016] FIG. 2 is a diagram showing process data detected by the system for controlling the quality of cold-rolled steel sheets according to an embodiment of the present invention.

[0017] Referring to FIG. 2, the process data shown in the figure is data obtained by measuring the yield strength (YP), tensile strength (TS), and elongation (EL) in the length direction from the head to the tail of the cold-rolled steel sheet.

[0018] As the learning data, elements that have a significant effect on the material quality are selected, and the component data can be retrieved for each coil of cold-rolled steel sheet, and data relating to the length (total length) of the hot rolling, cold rolling, and annealing processes can be stored in the first database 151. More specifically, referring to FIG. 2, the temperature just before hot finish rolling (FET), the temperature just after hot finish rolling (FDT), and the coiling temperature (CT) can be measured and collected, and such length-wise data can be used as learning data for the hot rolling process. As the hot rolling process data, data measured at predetermined intervals, for example, at equal intervals within 10 m, can be collected along the entire length of the coil.

[0019] By linking the hot rolling process data with the data from the subsequent cold rolling and annealing process, the consistency of the material prediction model can be improved compared to when each variable in the hot rolling process is used as an average value for each process, as shown in Table 1 below. In other words, by using data on material measurement sites as hot rolling-related variables rather than average values ​​over the entire length, the prediction consistency can be improved.

[0020] [Table 1]

[0021] FIG. 3 is a schematic flowchart of a method for generating a material quality prediction model for a cold-rolled steel sheet according to one embodiment of the present invention.

[0022] Referring to Figures 1 and 3, the model generator 153 generates a training data set based on the process data stored in the first database 151 by querying the process data stored in the first database 151 (S1), merging it with material data, and then generating a training data set (S2), and training a pre-set model to generate a predictive model (S3).

[0023] First, the step of generating a training dataset may include a step (S1a) of querying component data of a slab coil, a step (S1b) of querying the collected temperature just before hot finish rolling (FET), temperature just after hot finish rolling (FDT), and coiling temperature (CT), a step (S1c) of querying the roll force value during cold rolling of the coil, and a step (S1d) of querying the annealing temperature value for the entire length of the coil. Such entire length data can be used as training data as hot rolling data. Regarding the cold rolling-related data, data related to the roll force generated during cold rolling is measured and collected, and the steel sheet threading speed from the HS to the FCS in a continuous annealing furnace and in the annealing furnace is measured and collected. Because the coil dimensions change significantly between the hot rolling and cold rolling processes, a measurement position mapping process is required between the hot rolling data and the cold rolling / annealing data. This allows the hot rolling data to be expanded by the length of the cold rolling and then merged with the cold rolling and annealing data.

[0024] To explain the cold rolling-related data in more detail, first, data on roll force generated during cold rolling is measured and collected, and then the speed at which the steel sheet passes through the heating zone, the final cooling zone, and the annealing furnace is measured and collected. Because the lengths of the hot-rolled steel sheet and the cold-rolled steel sheet are different, a step of position mapping each of the hot-rolling / cold-rolling and annealing data can be further performed.

[0025] FIG. 4 is a diagram showing learning data variables in a method for generating a material quality prediction model for cold-rolled steel sheets according to one embodiment of the present invention, and FIGS. 5 and 6 are diagrams showing weight settings in a method for generating a material quality prediction model for cold-rolled steel sheets according to one embodiment of the present invention.

[0026] First, referring to FIG. 4 in conjunction with FIG. 1, there is a step of querying the model generator 153 for measured and collected data from the first database 151. This data is data related to process variables and corresponds to X data in the model. Output, i.e., data related to Y, is material measurement data and corresponds to yield strength, tensile strength, and elongation. For each coil, the process variables are stored in the second database 152. Information on newly produced coils is periodically stored in the second database 152 (for the variables mentioned above), and the period may be one week to one month. As described above, the model generator 153 is re-queried for the learning data stored in the second database 152, and outlier removal logic is applied to the retrieved data. The outlier removal logic has a function of removing data below or above the first or third quantile based on the commonly used 1.5*IQR (Interquartile Range) or filtering erroneous data during data collection.

[0027] As shown in Figure 4, training and test sets are generated separately for X, Y pairs in the generated training data, and model training is performed only on the training data. The trained model is evaluated using the test set, and the means for evaluating the consistency of the model may be RMSE (Root Mean Square Error) and MAPE (Mean Absolute Percentage Error).

[0028] 1 and 3, a training data set is formed by merging the generated data with material data (output variables), and then a preset model in the model generator 153 is trained. The model may be an ensemble model of a neural network algorithm consisting of three hidden layers and machine learning techniques such as XGBoost and Random Forest. The training data may be formed in the form of a table as shown in FIG. 4 for each length direction. Material values ​​may be obtained for each length direction, and models may be formed individually for each length direction using process data and material data for the length direction. A predictive model may be generated using the training data set (S3).

[0029] More specifically, this method uses training data to generate regression models with the XGBoost algorithm, Random Forest algorithm, and Neural Network algorithm, and then uses the average of the output values ​​derived (predicted) by each model as the final predicted value. When using such models, consistency is improved by more than 10% compared to conventional methods. More specifically, for a steel grade with a yield strength of 1000 MPa, the results are as shown in Table 2 below. (Example of yield strength prediction performance)

[0030] 5 and 6, along with FIGS. 1 and 2, model consistency can be improved by applying different weights to each model when ensembling the XGBoost algorithm, the Random Forest algorithm, and the Neural Network algorithm. The logic for this is to derive the weights for each model using the Gradient Boosting technique, and to assign arbitrary weights initially and train the models until the RMSE becomes low, as shown in FIG. 6. Here, if a value of 1 is assigned to each model when initializing the weights, the learning speed can be further improved.

[0031] [Table 2]

[0032] FIG. 7a is a flowchart showing the overall length material quality control of a cold-rolled steel sheet according to one embodiment of the present invention, and FIG. 7b is a diagram showing an example of presenting an optimal annealing temperature over the entire length in a cold-rolled steel sheet material quality control system according to one embodiment of the present invention.

[0033] 7a and 7b together with FIGS. 1 and 3, the model generator 153 includes a step (S4) of deriving an optimal annealing temperature for finding the target material using a particle swarm optimization (PSO) algorithm based on the predictive model.

[0034] That is, once the above-described prediction model is completed, the next step is to construct a model that derives the annealing temperature, one of the X variables, to achieve the target material quality based on the prediction model. More specifically, this method involves determining the annealing temperature to find the optimal material quality, with the composition, hot rolling variables, and cold rolling variables as fixed variables, and only the annealing temperature as a variable. The annealing temperature variable can be one or more variables. The algorithm then uses particle swarm optimization to find the optimal annealing temperature. The total length data from hot rolling to annealing used may be, for example, 100 to 2,000 pieces per coil (depending on the coil production length and production speed), and the optimal annealing instruction value is derived for each point (i.e., data) along the entire length of the coil. By setting different annealing temperatures according to the total length in the continuous annealing furnace, it is possible to reduce the material quality variation caused by the hot rolling process, i.e., the material quality variation over the entire length that occurs during the hot rolling process. By matching the material quality to the median value of the material specification, more stable material quality can be achieved. The annealing temperature value is given for each point along the entire length, as shown in Figure 7b. Once the optimum annealing temperature is derived in this manner, the value can be stored in the second database 152, and the optimum annealing temperature value stored in the second database 152 can be sent to the production control system 140. The optimum annealing temperature sent to the production control system 140 can be reflected in the factory design value.

[0035] In this operation, a prediction model and an optimum annealing temperature derivation model can be formed for each of points spaced apart by a predetermined unit having a preset distance in the length direction, for example, 5 to 15 points.

[0036] FIG. 8 is a diagram illustrating an exemplary computing environment in which a material property control system for cold-rolled steel sheets according to an embodiment of the present invention can be implemented.

[0037] 8 illustrates an example of a system 1000 including a computing device 1100 configured to implement a material quality control system for cold rolled steel sheets according to an embodiment of the present invention. For example, the computing device 1100 may include, but is not limited to, a personal computer, a server computer, a handheld or laptop device, a mobile device (such as a mobile phone, PDA, or media player), a multiprocessor system, a consumer electronics device, a minicomputer, a mainframe computer, a distributed computing environment including any of the aforementioned systems or devices, and the like.

[0038] The computing device 1100 may include at least one processing unit 1110 and memory 1120. Here, the processing unit 1110 may include, for example, a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), etc., and may have multiple cores. The memory 1120 may be volatile memory (e.g., RAM, etc.), non-volatile memory (e.g., ROM, flash memory, etc.), or a combination thereof.

[0039] Computing device 1100 may also include additional storage 1130, including but not limited to magnetic storage, optical storage, etc. Storage 1130 may store computer-readable instructions for implementing at least one embodiment disclosed herein, as well as other computer-readable instructions for implementing an operating system, application programs, etc. The computer-readable instructions stored in storage 1130 may be loaded into memory 1120 for execution by processing unit 1110.

[0040] The computing device 1100 may also include input devices 1140 and output devices 1150. Here, the input devices 1140 include, for example, a keyboard, a mouse, a pen, a voice input device, a touch input device, an infrared camera, a video input device, or any other input device. The output devices 1150 include, for example, one or more displays, speakers, a printer, or any other output device. The computing device 1100 may also use input devices or output devices provided in other computing devices as the input devices 1140 or the output devices 1150.

[0041] Computing device 1100 may also include communications connection(s) 1160 that allow it to communicate with other devices (e.g., computing device 1300) over a network, such as the Internet 1200. Here, communications connection(s) 1160 may include a modem, a network interface card (NIC), an integrated network interface, a radio frequency transmitter / receiver, an infrared port, a USB connection, or other interface for connecting computing device 1100 to other computing devices. Also, communications connection(s) 1160 may include wired or wireless connections.

[0042] The components of the computing device 1100 described above may be connected by various interconnections such as buses (e.g., Peripheral Component Interconnect (PCI), USB, firmware (IEEE 1394), optical bus structures, etc.), or may be interconnected by a network.

[0043] As used herein, terms such as "programmable logic controller (PLC)," "process controller (PC)," and "manufacturing execution system (MES)" generally refer to computer-related objects that are hardware, a combination of hardware and software, software, or software in execution. For example, a component of a "programmable logic controller (PLC)," "process controller (PC)," or "manufacturing execution system (MES)" may be, but is not limited to, a process running on a processor, a processor, an object, an executable, a thread of execution, a program, and / or a computer. For example, both an application running on a controller and the controller may be a component. One or more components may reside within a process and / or thread of execution, and a component may be localized on one computer or distributed across two or more computers.

[0044] As described above, the present invention utilizes an artificial intelligence model, enabling a wider range of data and a larger amount of data than conventional physical models, thereby demonstrating more precise and accurate prediction capabilities. Furthermore, the online model configuration allows the system to operate continuously from data generation to calculation, supporting the realization of smart factories. In addition, for cold-rolled products, although the hot rolling process is the first process, data for each position within the coil for all processes from hot rolling to cold rolling is matched to the through-die, providing a more accurate prediction model. Furthermore, with the above-described configuration of the present invention, material information for the entire length can be obtained, as shown in Figure 2, and the cause of material differences at different positions within the coil can be identified. Providing such full-length material data can address factors that reduce production capacity during forming, such as mold modifications due to plate material portions within the coil and changes in forming conditions.

[0045] The present invention described above is not limited to the above-mentioned embodiments and the accompanying drawings, but is limited by the claims, and it will be easily understood by those having ordinary knowledge in the field of technology to which the present invention pertains that the configuration of the present invention can be changed and modified in various ways within the scope that does not deviate from the technical idea of ​​the present invention.

Claims

1. A programmable logic controller that acquires process data for cold-rolled steel sheets that have already been produced; a process controller that controls a manufacturing process of the cold-rolled steel sheet according to the process data of the programmable logic controller; a model generation unit that learns a learning dataset in advance and generates a prediction model that predicts the material properties of the cold-rolled steel sheet in the longitudinal direction to be produced; a production management system that transmits and receives information to and from the process controller, receives the predicted value of the prediction model generated by the model generation unit, and manages the production of the cold-rolled steel sheet; Including, The training data set is Hot rolling process data among the process data from the programmable logic controller; Material data and component data from the production management system; data mapping measurement positions between cold rolling and annealing process data among the process data from the programmable logic controller; A material quality control system for cold rolled steel sheets, including:

2. 2. The system for controlling the material quality of cold-rolled steel sheets according to claim 1, wherein the model generation unit performs data extension on the hot rolling process data by the length extended by cold rolling, merges the data with the cold rolling and annealing process data, and corrects the learning data set.

3. 2. The system for controlling the material properties of cold-rolled steel sheets according to claim 1, wherein the data acquired by the programmable logic controller includes material property data measured in predetermined units in the length direction of the cold-rolled steel sheets.

4. The system for controlling the material quality of a cold-rolled steel sheet according to claim 3 , wherein the model generation unit generates the prediction model for each of the predetermined units for measuring the material quality data in the length direction of the cold-rolled steel sheet.

5. 2. The system for controlling the material quality of a cold-rolled steel sheet according to claim 1, wherein the model generation unit further generates an annealing temperature derivation model that derives an optimal annealing temperature for finding a target material quality using a preset particle swarm optimization (PSO) algorithm based on the prediction model, and the production management system controls the process controller according to the optimal annealing temperature of the annealing temperature derivation model to control the annealing temperature of the cold-rolled steel sheet.

6. 2. The system for controlling a material quality of a cold-rolled steel sheet according to claim 1, wherein the model generation unit generates the prediction model including a preset XGboost algorithm, a Random Forest algorithm, and a neural network algorithm, and outputs a final prediction value by averaging prediction values ​​of the XGboost algorithm, the Random Forest algorithm, and the neural network algorithm of the prediction model.

7. 7. The system for controlling a material quality of a cold-rolled steel sheet according to claim 6, wherein the model generation unit weights each predicted value of the XG boost algorithm, the random forest algorithm, and the neural network algorithm of the prediction model by a preset gradient boosting.

8. a step in which a model generation unit generates a training dataset; generating a prediction model for predicting the longitudinal properties of the cold-rolled steel sheet to be produced by training a pre-established model on the training data set; Including, The plurality of data included in the training dataset are The data includes hot rolling process data among the process data provided from a programmable logic controller, material data and component data provided from a production management system, and data mapping measurement positions between cold rolling and annealing process data among the process data from the programmable logic controller, The process data provided by the programmable logic controller includes: A method for generating a material quality prediction model for cold-rolled steel sheets, which is process data for produced cold-rolled steel sheets acquired in advance by sensors.

9. The step of generating the training data set comprises:

9. The method for generating a material property prediction model for a cold-rolled steel sheet according to claim 8, wherein the model generation unit performs data extension on the hot rolling process data by a length elongated by cold rolling, merges the data with the cold rolling and annealing process data, and corrects the learning data set.

10. In the step of generating the training data set, 9. The method for generating a material quality prediction model for a cold-rolled steel sheet according to claim 8, wherein the data acquired by the programmable logic controller from the sensor includes material quality data measured in predetermined units in the length direction of the cold-rolled steel sheet.

11. 11. The method of claim 10, wherein in the generating the prediction model, the model generating unit generates the prediction model for each of the predetermined units for measuring the material data in a length direction of the cold-rolled steel sheet.

12. 10. The method of claim 8, further comprising: generating an annealing temperature derivation model for deriving an optimal annealing temperature for finding a target material using a preset particle swarm optimization (PSO) algorithm based on the prediction model; and controlling the production management system to control a process controller according to the optimal annealing temperature of the annealing temperature derivation model to control the annealing temperature of the cold-rolled steel sheet.

13. 9. The method of claim 8, wherein in the generating the prediction model, the model generating unit generates the prediction model including a preset XGboost algorithm, a Random Forest algorithm, and a neural network algorithm, and outputs a final prediction value by averaging prediction values ​​of the XGboost algorithm, the Random Forest algorithm, and the neural network algorithm of the prediction model.

14. 14. The method of claim 13, wherein in the generating the prediction model, the model generating unit weights the predicted values ​​of the XG Boost algorithm, the Random Forest algorithm, and the Neural Network algorithm of the prediction model by a preset gradient boosting.

Images