WO2019124931A1

WO2019124931A1 - Furnace condition control apparatus and method

Info

Publication number: WO2019124931A1
Application number: PCT/KR2018/016113
Authority: WO
Inventors: 한경룡; 이진휘; 손상한; 손기완
Original assignee: 주식회사 포스코
Priority date: 2017-12-19
Filing date: 2018-12-18
Publication date: 2019-06-27
Also published as: JP7050934B2; KR20190074132A; EP3730630A4; CN111492070A; KR102075210B1; EP3730630B1; JP2021507115A; EP3730630A1

Abstract

The present invention relates to a furnace condition control apparatus and method for guiding preemptive actions to stably maintain the condition of a blast furnace, by using various operational data and sensor data generated from the blast furnace. The furnace condition control apparatus according to an embodiment of the present invention may comprise: a first sensor part for imaging at least one of temperature data and pressure data of a blast furnace according to measurement positions; a second sensor part for detecting unstructured data of the blast furnace; and an action guidance part having an artificial intelligence algorithm for outputting an action guidance for operating the blast furnace, on the basis of the imaged temperature data or pressure data from the first sensor part and the unstructured data from the second sensor part. The furnace condition control method according to an embodiment of the present invention may comprise the steps of: collecting, by a data preprocessing part, unstructured data of at least one of the feed material condition, the tuyere condition, and the taphole condition of a blast furnace, and imaging temperature data and pressure data of the blast furnace according to measurement positions; receiving the data preprocessed by the data preprocessing part and outputting an action guidance for operating the blast furnace, by an artificial intelligence algorithm; determining whether the artificial intelligence algorithm requires relearning, according to whether an operator employs the action guidance; and determining whether to replace the artificial intelligence algorithm, according to whether the corresponding artificial intelligence algorithm achieves the relearning.

Description

Apparatus and method for managing agoraphobia

The present invention relates to an apparatus and method for managing an agar apparatus for managing an agar apparatus for blast furnace.

The blast furnace process is a representative process that mainly conducts the manual operation depending on the experience and intuition of the operator so far during the steel making process.

The blast furnace is a facility that charges iron ore and coke to the upper part of the blast furnace, blows hot air through the blast furnace tuyere, and produces liquid iron by the oxidation and reduction reaction inside through outlet. Because the inside of the blast furnace can not be measured through the sensor due to high temperature and high pressure, the condition of the blast furnace is predicted indirectly through the thermometer and pressure gauge installed on the outer wall of the blast furnace.

There are various indicators of the present condition of blast furnace, ie, aging. Three representative examples are the ventilation, breathability, and circumferential balance. In the case of furnace heat, it refers to an index for predicting the temperature inside the blast furnace through manual measurement of the temperature of the boiler coming out through the outlet. In the case of air permeability, the state of the hot air moving from the lower part to the upper part inside the blast furnace is indirectly , And the circumference balance is an index for a situation in which a circular blast furnace does not have a large difference in pressure and temperature in the circumferential direction, that is, the balance is maintained.

In order to keep these three indicators at the desired value, the operator takes action. Typical examples include control of pulverized coal (PCI) injection amount, control of air volume of hot wind, control of oxygen amount in air volume, ratio control of iron oxide and coke to be charged, and distribution control of large particle size coke in the center part.

Currently, the blast furnace operation basically judges the condition of the blast furnace according to the experience of the operator and intuition and the operating standards based on the form data such as the thermometer or the pressure gauge measured value and the information obtained through the atypical data such as CCTV, Based on this, we are taking action.

However, for more stable management of aging, it is important to anticipate how aging will take place through the current situation and current action, and to carry out the operation.

Such prior art can be easily understood with reference to Korean Patent Laid-Open Publication No. 10-1995-0014631.

According to an embodiment of the present invention, there is provided an apparatus and method for controlling a sulfur content management system for guiding pre-existing actions for stably maintaining the sulfur content using various operational and sensor data generated in the blast furnace.

According to an aspect of the present invention, there is provided an apparatus for managing an anther of a certain type, comprising a first sensor unit for imaging at least one of temperature and pressure data of a blast furnace according to a measured position, An action guidance having an artificial intelligence algorithm for outputting an action guidance on blast furnace operation based on the imaged temperature or pressure data from the first sensor unit and the atypical data from the second sensor unit; Section.

According to an embodiment of the present invention, there is provided a method for managing a glaucophore according to an embodiment of the present invention, which comprises collecting at least one irregular data of a charge condition, a trough condition and an exit condition of a blast furnace to a blast furnace, A step of inputting the preprocessed data of the artificial intelligence algorithm and outputting an action guidance related to the blast furnace operation, a step of judging re-learning of the artificial intelligence algorithm according to whether the action guidance of the operator is applied or not, And determining the replacement of the artificial intelligence algorithm according to re-learning.

According to an embodiment of the present invention, it is possible to stably produce blast furnace, improve blast furnace efficiency, maintain lukewarm performance maintaining constant performance, and automate and standardize operation.

1 is a schematic block diagram of an apparatus for managing the effect of the present invention according to an embodiment of the present invention.

FIG. 2 is a view for explaining the concept of artificial intelligence applied to an apparatus for managing aged people according to an embodiment of the present invention.

3 is a schematic operation flow diagram of a method for managing an aged population according to an embodiment of the present invention.

4 is a diagram showing an example of a GUI of a management apparatus for an aged person according to an embodiment of the present invention.

FIG. 5 is an image of a thermometer and pressure gauge data applied to a glaze management apparatus according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings, in order that those skilled in the art can easily carry out the present invention.

Referring to FIG. 1, an apparatus 100 for managing an aged person according to an exemplary embodiment of the present invention may include a first sensor unit 110, a second sensor unit 120, and an action guidance unit 130.

The first sensor unit 110 may image at least one of the temperature and pressure data of the blast furnace according to the measured position.

The first sensor unit 110 may include a temperature sensor unit 111, a pressure sensor unit 112, and a data processing unit 113.

The temperature sensor unit 111 may include a plurality of temperature sensors respectively installed in the blast furnace, and the plurality of temperature sensors can detect the temperature of the blast furnace at the installed position.

The pressure sensor unit 112 may include a plurality of pressure sensors respectively installed in the blast furnace, and the plurality of pressure sensors can detect the pressure in the blast furnace at the installed position.

The data processing unit 113 can map the detected temperature data of each of the plurality of temperature sensors of the temperature sensor unit 111 to the detected position and image them. Similarly, the detected pressure data of each of the plurality of pressure sensors of the pressure sensor unit 112 can be mapped and imaged to the detected position. In addition, the detected temperature data of each of the plurality of temperature sensors and the detected pressure data of each of the plurality of pressure sensors can be mapped and imaged to the detected position.

Due to the nature of the blast furnace, there may be a cross-correlation between temperature and pressure. Therefore, if it is used as the input data of the deep learning algorithm, it is advantageous to analyze the blast condition, and it can be a main factor for improving the performance for action guidance.

The data processing unit 113 may map the detected temperature or pressure data to the detected position to form a two-dimensional image.

Referring to FIG. 5 together with FIG. 1, an example in which sensor data of blast furnace, that is, detection data of the temperature sensor unit 111 and the pressure sensor unit 112 are imaged.

5, it is assumed that a plurality of temperature sensors are distributed on the surface of a column-shaped blast furnace, and a heat map is created and cut out at 0 degree. That is, the horizontal direction of the drawing is the angle at which the temperature sensor is distributed. And the height distribution is corresponding to the height of the drawing. As a result, each black dot represents a temperature sensor. As can be seen from the left and center figures, the temperature values of the blast furnace vary instantaneously with an organic correlation.

In the case of the pressure sensor shown in the right figure, a typical four-direction value is expressed. Directional pressure gauges can be divided into four color lines. The horizontal axis represents the pressure value, and the vertical axis represents the height position of the pressure sensor. In the present invention, the imaging technology as shown in this drawing is used to efficiently input the necessary positional information relation to the artificial intelligence.

The second sensor unit 120 can detect atypical data of the blast furnace by measuring at least one of the state of the blast furnace, the tuyere state, and the outlet state.

According to the present invention, it is possible to determine an aging based on current blast state data through an algorithm based on a deep learning and to suggest an optimal action guidance for maintaining a normal aging state. Since deep-run-based algorithms are data-driven algorithms, a large amount of data is necessary to represent the situation well.

Therefore, the operators have used the data as the basis of the judgment of the blind operation by the naked eye, but the data that can not be used for the control using the computer because of the unstructured data is formulated and applied to the present invention.

The first data is data on the particle size of the iron ore and coke charged. This is data related to breathability.

The second is to use the data of the combustion zone of the tungsten as a numerical data. In the case of tuyu burning zone, it is the only part that can observe inside the blast furnace, and it is a facility to blow hot air. Here, the pulverized coal is blown in together, and it functions to monitor the combustion state of the pulverized coal and the combustion / raw material that does not melt at the inner wall of the blast furnace.

The third is an instrument for the exit condition, and in particular, the measurement of the char temperature is an important factor. For the basic blast furnace operation, measure the temperature of the incoming charcoal manually once every 1 to 2 hours. The measurement position is also a place at a distance from the exit, and the degree of measurement of the person is not constant. This value is important data related to the row.

To this end, the second sensor unit 120 may include a charge state meter 121, a tougue state meter 122, and an exit state meter 123.

The charge state meter 121 can measure at least one of the particle size, particle size distribution, and humidity state of the charge placed in the conveyor belt passing through the soft material to be charged into the blast furnace of the blast furnace, and measures the measured unstructured data as the format data And transmits it to the action guidance unit 130.

The toughening state measuring instrument 122 can measure at least one of the pulverized coal injection state and the dropping state of the blast furnace through the plurality of toughen cameras, converts the measured irregular data into the formatted data, and transmits the formatted data to the action guidance unit 130 .

The exit meter status measuring device 123 measures the temperature of the molten iron leaving the blast furnace in real time and measures the amount of emission or the like based on the angle and thickness of the molten steel stem and converts the measured irregular data into the formatted data, 130).

The action guidance unit 130 may output the action guidance related to the blast operation based on the imaged temperature or pressure data from the first sensor unit 110 and the atypical data from the second sensor unit 120. [

Referring to FIG. 2 together with FIG. 1, the action guidance unit 130 may include a learning unit 131, a control unit 132, and a reinforcement learning unit 133.

Referring to FIG. 3 together with FIGS. 1 and 2, the learning unit 131 may include an action guidance on-line algorithm, and the action guidance on-line algorithm may include a temperature and pressure from the first sensor unit 110, Based on the data S10 and S11 and the state of the charge of the blast furnace formulated from the second sensor unit 120, the toughening state and the exit state state measurement data S10 and S12, and the action guidance (S20, S30).

The action guidance online algorithm is composed of a deep learning-based algorithm,

) To learn the action guidance (

Can be generated.

The control unit 132 receives the action guidance of the learning unit (

, And whether or not the operator's action guidance is accepted can be fed back to the reinforcement learning unit 133 (S40).

The reinforcement learning unit 133 may include an action guidance offline algorithm configured by a deep learning based algorithm, and the action guidance offline algorithm may enhance the algorithm learning by receiving an action guidance not accepted by the worker.

The control unit 132 may determine whether to re-learn the action guidance online algorithm and replace the action guidance online algorithm with the action guidance offline algorithm of the reinforcement learning unit 133. [

In other words, deep irregular data that is generated in the blast furnace and important for the operation, and existing formal data are collected and inputted into the artificial intelligence system that utilizes deep learning, the deep learning algorithm is operated based on the learned model Provide guidance on the action to be performed by the user. The operator judges acceptance of such action guidance, and the deep learning algorithm uses it as feedback to utilize the algorithm to enhance the performance. In addition, if the characteristics of the input data are changed after a certain period of time or more than a predetermined standard, re-learning is performed to maintain the artificial intelligence algorithm for the current blast condition to optimize the performance.

More specifically, the collected data is preprocessed by combining the input of the unstructured data with the standardized data and the input of the fixed data, and the data is input to the deep learning-based action guidance online algorithm. Here, the algorithm presents action guidance based on its model. The proposed action guidance value is judged by the operator to be acceptable for operation and accepted or rejected. Through this loop, the first algorithm is used.

In addition, the on-line learning or reinforcement learning is performed by receiving as a feedback value a result of whether the operator accepts the AI action guidance when there is an offline offline control algorithm (S60). That is, in the case of the deep learning-based action guidance offline (off-line) algorithm, the action guidance value inputted as the feedback according to the acceptance of the former operator is compensated (S50) and used for the algorithm reinforcement. Fundamentally, the deep learning-based action guidance offline algorithm has a reinforcement learning part, which is used to improve the algorithm performance in case of misidentification of the deep learning-based action guidance on-line algorithm. In addition, if the compensation value falls below a predetermined level or the characteristics of the data are learned, the re-learning is judged and the re-learning is performed if necessary (S70).

If the re-learning result requires algorithm replacement (S80), the deep learning-based action guidance on-line algorithm is replaced with the newly learned action guidance offline algorithm on the system. Therefore, it is possible to maintain the algorithm that responds to the blast situation, and the more the operation is performed, the better the action guidance performance can be realized.

FIG. 4 is a diagram showing an example of a GUI (Graphic User Interface) of an apparatus for managing the aged population according to an embodiment of the present invention.

Referring to FIG. 4 together with FIG. 1, the action guidance unit 130 can present actions related to blast furnace operation such as air volume, oxygen, pulverized coal, loading / raw material cost, and distribution of center coke. For example, through the illustrated GUI, an action guidance value necessary for air volume control can be confirmed and the trend of related data can be confirmed. If necessary, manual operation can also be carried out.

As described above, according to the present invention, it is possible to produce stable blast furnace by guiding the action of the operator who needs to maintain a stable sulfur content, thereby improving the efficiency of the blast furnace. In addition, it is possible to maintain the longevity management system with a certain performance by keeping the algorithms capable of coping with the operating conditions and the blast conditions varying with time. In addition, since the operation is automated and standardized, it is possible not only to reduce the load of the operator, but also to formulate the tactile know-how and experiential experience such that the tactic can be shared with the propagation.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not intended to limit the invention to the particular forms disclosed. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

A first sensor unit for imaging at least one of temperature and pressure data of the blast furnace according to the measured position;

A second sensor unit for detecting irregular data of the blast furnace; And

An action guidance unit having an artificial intelligence algorithm for outputting an action guidance on blast furnace operation based on the imaged temperature or pressure data from the first sensor unit and the atypical data from the second sensor unit,

And a control unit.
The method according to claim 1,

The first sensor unit

A temperature sensor unit having a plurality of temperature sensors for measuring the temperature of each position of the blast furnace;

A pressure sensor unit having a plurality of pressure sensors for measuring pressures at respective positions of the blast furnace; And

A data preprocessing unit for matching the measured temperature and pressure of each of the temperature sensor unit and the pressure sensor unit with the measured position,

And a control unit.
3. The method of claim 2,

Wherein the data preprocessor matches the measured temperature and pressure with the measured position and implements the image in two dimensions.
The method according to claim 1,

And the second sensor unit measures at least one of the state of the charge in the blast furnace, the state of the tuyere, and the state of the outlet port.
5. The method of claim 4,

The second sensor unit

A charge state meter for measuring at least one of particle size, particle size distribution and humidity state of the charge in the blast furnace;

A tundish state meter for measuring at least one of a pulverized coal blowing state and a live light falling state of the blast furnace; And

And an output port state meter for measuring at least one of a molten iron temperature and an output of the blast furnace

And a control unit.
The method according to claim 1,

And the second sensor unit converts the collected irregular data into the formatted data and transmits the converted data to the action guidance unit.
The method according to claim 1,

The action guidance section

A learning unit having an action guidance online algorithm for learning based on data collected from the first sensor unit and the second sensor unit and generating an action guidance related to blast furnace operation;

An enhancement learning unit having an action guidance offline algorithm for enhancing algorithm learning according to acceptance of the action guidance of the operator; And

And a control unit for outputting the action guidance of the learning unit and re-learning of the action guidance on-line algorithm and judging whether the action guidance on-line algorithm is replaced with the action guidance offline algorithm of the reinforcement learning unit

And a control unit.
Collecting at least one irregular data of the charge state, the tidal state, and the exit state of the data preprocessing section to the blast furnace, and imaging the blast furnace temperature and pressure data according to the measured position;

Outputting an action guidance related to the blast furnace operation by receiving the preprocessed data from the artificial intelligence algorithm;

Determining re-learning of the artificial intelligence algorithm according to whether the action guidance of the agent is applied; And

A step of judging replacement of the artificial intelligence algorithm according to the re-learning of the artificial intelligence algorithm

/ RTI >