JP3144984B2 - Adjustment method of molten steel temperature in steelmaking process - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for adjusting the temperature of molten steel in a steelmaking process for adjusting the temperature of molten steel in its own process based on a request from a subsequent process.

[0002]

2. Description of the Related Art A steelmaking process includes a converter process, a vacuum degassing process such as RH and DH, and a continuous casting process. Molten steel that has exited the converter process is continuously cast through a vacuum degassing process. There is a case 1 that reaches the process or a case 2 where the molten steel that has exited the converter process directly reaches the continuous casting process.

For example, as shown in FIG. 2, the temperature of the molten steel in the case 1 is radiated during transportation or processing.
It descends sequentially due to heat removal by the molten steel pot and tundish.

On the other hand, in the continuous casting step, which is the last step, since the casting temperature determines the quality of the cast slab, it is most important to keep the casting temperature within a specified value. For this reason, taking into account the temperature drop of the molten steel caused by heat removal from the molten steel pot or heat radiation during the movement of the molten steel pot, the casting temperature is first adjusted before the continuous casting process so that the casting temperature is within the above-mentioned specified value. The temperature in the vacuum degassing step, which is a step, is set, and then the tapping temperature in the converter step, which is the preceding step, is determined based on the set temperature in the vacuum degassing step. That is, this temperature setting reverses the unsteady heat conduction phenomenon in consideration of the heat removal and heat radiation of the molten steel pot from the continuous casting step which is the last step.

For this reason, for example, Japanese Patent Application Laid-Open No. HEI 3-161161
As shown in the publication, a model formula consisting of a refractory front and back surface temperature of a refractory lining of a molten steel pot that has been emptied by dispensing molten steel, a specific heat and weight of the refractory, and a heat radiation correction coefficient obtained from an experiment. First, to determine the heat storage amount of the molten steel pot and the refractory lining of the tundish, and then to predict the temperature drop of the molten steel using the heat storage amount, furthermore, to know the temperature drop of the molten steel after a certain period of time Measure the molten steel temperature at least twice
Using this measured value, the error of the heat storage amount of the refractory calculated by the model formula is corrected, and the subsequent molten steel temperature drop is re-predicted. Based on the predicted molten steel temperature drop, the setting of the molten steel temperature in the previous process is performed. There is a way to do it.

[0006]

However, the method disclosed in JP-A-3-161161 has the following problems in prediction accuracy, reliability of temperature measurement values, and calculation load. That is, since the refractory temperature, weight, and the like that constitute the above model formula for estimating the heat storage amount of the molten steel pot have large error factors as described in the following, creating a simplified model that takes these factors into account is extremely heavy. Becomes Also,
Even if a model is made, the maintenance for maintaining its accuracy is enormous, and it is not practical.

[0007] The surface temperature and the back surface temperature (often replaced by the outer wall steel temperature) of a refractory lining a molten steel pot are measured by a radiation thermometer or a contact thermocouple thermometer.
This contact type thermocouple requires much time for measurement, and the radiation thermometer causes an error of several tens of degrees C depending on the surface condition of the molten steel pot.

[0008] The temperature of the refractory lining of the molten steel pot is not uniform throughout the molten steel pot, and a temperature distribution of about several tens to 100 ° C is generated, and the error due to the representativeness of the measurement points becomes large.

As the number of uses of the molten steel pot increases, the weight of the refractory lining decreases, but the rate of change is not uniform, and the weight must be measured frequently, which is very troublesome and requires an error. Also occurs.

[0010] Since the properties of the refractory also change as it comes into contact with molten steel, the specific heat of the refractory also changes, causing an error.

It is conceivable to use a higher-order nonlinear differential equation in order to set the molten steel temperature in the preceding process and to solve this by convergence calculation. For example, when setting the temperature of the vacuum degassing process, which is the preceding process from the continuous casting process, first,
The temperature in the vacuum degassing process is provisionally set, the temperature is substituted into the model to calculate the temperature in the continuous casting process, and the temperature in the temporarily set vacuum degassing process is corrected by comparing with the prescribed casting temperature. This is a convergence calculation in which the temperature of the vacuum degassing step when the specified casting temperature is reached by repeating this is a set value.
This requires a large computational load and requires a high-speed computer, which increases costs, and makes it difficult to quickly respond to changes in casting temperature when responding to operational fluctuations. There was no one.

For this reason, this temperature setting operation is performed based on the intuition and experience of a skilled operator. However, this setting cannot be performed with sufficient accuracy due to the individual difference of the skilled operator. There were problems such as the need to constantly train and maintain excellent skilled operators.

The present invention has been proposed in view of such circumstances, and based on requirements from a post-process, the temperature of molten steel in its own process can be accurately and accurately determined without using a skilled operator using a hierarchical neural network. The purpose is to seek quickly and automatically.

[0014]

Means for Solving the Problems In a steelmaking process in which molten steel is received in a molten steel ladle from a converter process and the molten steel is conveyed to a continuous casting process directly or through a vacuum degassing process, the converter process is performed by itself. When the vacuum degassing process is a post-process, when the vacuum degassing process is a self-process and the continuous casting process is a post-process, and further, when the converter process is a self-process and the continuous casting process is a post-process, The number of times the molten steel ladle has received a certain amount of molten steel from the converter process, how long the molten steel has been in the molten steel ladle before receiving the molten steel, and The empty time of the molten steel pot before receiving the molten steel, the transport time required to transport the molten steel pot from the self-process to the subsequent process, and further, when the subsequent process is a continuous casting process, the molten steel of the charge is received. Continuous use of tundish When the number of times and the post-process are vacuum degassing processes, the actual data of the temperature drop factor of the molten steel such as the amount of alloy injected into the molten steel in the converter process and the data such as the temperature of the molten steel when the molten steel arrives in the post-process. , Input to the input layer of the hierarchical neural network, input molten steel temperature actual data in the self-process of the charging, and obtain the weighting factor and threshold value in the hierarchical neural network storing the respective actual data. The target temperature in the post-process of molten steel of the charge to be temperature-adjusted and the expected number of use of the molten steel pot used to receive the molten steel of the target charge in the input layer of the neural network based on the obtained weighting coefficient and the threshold value. , The time during which the previously received molten steel was present in the ladle, the scheduled time for the molten steel ladle to receive the molten steel of the target charge, Scheduled transfer time in the process, furthermore, the number of continuous use of the tundish when the post-process is a continuous casting process, the amount of alloy to be put into the molten steel in the converter process when the post-process is a vacuum degassing process, etc. The molten steel temperature of the target charge in the own process is determined by inputting the molten steel temperature drop factor data of the above step, and the temperature adjusting means adjusts the molten steel of the target charge to the determined molten steel temperature in the own process.

[0015]

The present inventor analyzed operating data in the steelmaking process, extracted factors affecting the molten steel temperature, and examined the difficulty and accuracy (reliability) of measuring the factors. As a result, if the vacuum degassing process is a self-process and the continuous casting process is a post-process, it can be easily measured without the need for a special sensor, and the data with less disturbance is used. When the continuous casting process is a post process, and when the converter process is its own process and the vacuum degassing process is a post process, (1) the heat storage state of the molten steel pot is common. As data: Number of times the molten steel pot has been used (number of times the molten steel has been received since the refractory lining was performed), and idle time of the molten steel pot (the time from when the previously received molten steel was discharged to when it received the molten steel) The time during which the previously received molten steel was present in the molten steel pot (the time from receiving the molten steel from the converter process to discharging the molten steel to the tundish in the continuous casting process) was selected.

(2) The time required for moving the molten steel pot containing molten steel (matching time) was selected as data representative of the state of heat release from the molten steel pot.

Further, when the vacuum degassing step is the self-process and the continuous casting step is the subsequent step, or when the converter step is the self-process and the continuous casting step is the subsequent step, the following are specific features: As the data representative of the state of heat storage in the tundish, the number of continuous uses was selected.

When the converter step is a self-process and the vacuum degassing step is a subsequent step, the following are specific: (4) Data representing temperature fluctuation due to a chemical reaction include:
The amount of alloy input to molten steel in the converter process was selected.

On the other hand, a three-layer neural network that excels in releasing higher-order nonlinear differential equations that determine the temperature of the molten steel in the own process based on the temperature of the molten steel required from the subsequent process was selected.

First, the results of (1) to (4), (2), (3) and (4) of the charge having a past casting record in the input layer of the three-layer neural network. Data, and the actual temperature data of the molten steel that has reached the post-process is input, and the actual data of the molten steel temperature of the charge in the own process is input and learned as teacher data,
The input layer, the intermediate layer,
Each weight coefficient and threshold value of the output layer are obtained, and based on the obtained weight coefficient and threshold value, the scheduled data of (1) to (2), (3) of the charge to be processed this time. By inputting the scheduled data of (4) and the target molten steel temperature data of the subsequent process to the input layer, the molten steel temperature of the own process is calculated. By adjusting the molten steel temperature in its own process so as to be the molten steel temperature calculated in this way,
The temperature of the molten steel that has reached the subsequent process is set to the target value.

As a result of the adjustment of the molten steel temperature in this way, when the molten steel reaches the post-process, the temperature of the molten steel becomes 9%.
5% or more became acceptable.

[0022]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 6 shows a configuration example of a molten steel temperature setting system in the present embodiment. In FIG. 6, reference numeral 1 denotes a higher-level computer, which sets a molten steel component adjustment target value and a temperature target value in each step in the steel making process. It is for. 2 is a converter processor (converter process control computer), 3 is a vacuum degassing computer (vacuum degassing process computer), 4
Is a continuous casting process computer (continuous casting process control computer), and each of the computers 2 to 4 is a computer for controlling a molten steel component, a temperature, and the like in each process based on a target value of the host computer 1.

FIG. 1 shows a process of the host computer 1 for setting a target temperature value. According to this flowchart, an example of the temperature setting in the vacuum degassing step based on the casting temperature required in the continuous casting step, which is one embodiment of the present invention, will be described.

First, in the host computer 1, as shown in FIG. 7, an input layer A having six neurons, an intermediate layer B having 50 neurons, and one
A three-layer neural network composed of the output layers C is formed (S1). Next, this host computer 1
In addition, the temperature of the molten steel when the molten steel discharged from the converter process reaches the vacuum degassing process, the number of times the molten steel ladle has received the molten steel, , The empty pot time, which is the empty time of the molten steel pot from the discharge of the molten steel from the pre-rotating furnace process to the reception of the molten steel, and the transfer of the molten steel from the vacuum degassing process to the continuous casting process. Required matching time, the required time, the number of continuous use of tundish in the continuous casting process,
And the actual casting temperature data for the tundish are collected and accumulated (S2).

Then, each of the above-mentioned data is inputted one by one to each neuron of the input layer A of the neural network, and the inputted data is substituted into Xj of the following equation (1). It is converted to Ui, which is the internal state of each neuron.

[0026]

(Equation 1)

Here, Wij is a weighting coefficient from the input layer A to the intermediate layer B, and a random number from -1 to 1 is substituted as an initial value.

Further, this Ui is calculated by the following equation (2).
To the output Yi of each neuron.

Yi = f (Ui + θi) (2) Here, θi is a threshold value different for each neuron of the intermediate layer B, and a random number from −1 to 1 is substituted as an initial value. F is a conversion function, and a sigmoid function shown in FIG. 5 is used.

The output Yi of each neuron in the intermediate layer B is converted into U, which is the internal state of the neuron in the output layer C, by the following equation (3).

[0031]

(Equation 3)

Here, Wj is a weight coefficient from the intermediate layer B to the output layer C, and a random number from -1 to 1 is substituted as an initial value.

Next, U which is the internal state of the neuron of the output layer C is converted into Y which is the output of the neuron of the output layer C by the following equation (4).

Y = f (U + θ) (4) Here, θ is a threshold value, and a random number from −1 to 1 is substituted as described above.

Here, the above-mentioned data as the teacher data d, that is, the actual data of the molten steel temperature in the vacuum degassing process which is its own process is given to the output layer C, and the output of the neuron of the output layer C is given to this teacher data d The weighting factors Wij, Wj and the threshold values θi, θi of the input layer A to the intermediate layer B and of the intermediate layer B to the output layer C of the neural network,
Is corrected according to the learning methods shown in the following (1) to (4).

(1) A learning coefficient δ is calculated by the following equation (6): δ = (d−Y) fa (U) (6) where fa is a differential value of the conversion function f. .

(2) Based on the calculated learning coefficient δ, the following
According to equation (5), a new weight coefficient Wi from the intermediate layer B to the output layer C is obtained.
(new) is calculated; Wi (new) = η · δ · Yi + Wi (old) (5) where i: 1 to 50, Wi (old): weight coefficient before learning, η: relaxation coefficient of learning .

(3) A learning coefficient δj is calculated by the following equation (7): δj = fa (Uj) · (δ · Wj) (7) (4) Based on the calculated learning coefficient δj, A new weight coefficient Wij (new) of the intermediate layer B is calculated from the input layer A by the equation (8); Wij (new) = η · δj · Xi + Wij (old) (8) where i: 1 to 6, j: 1 to 50.

In a similar manner, the threshold value θ of each neuron is corrected to obtain a new threshold value θ (new).

Next, based on the corrected new weighting factors Wi (new) and Wij (new) and the threshold value θ (new),
The calculation is performed again by each neuron of each input layer A, intermediate layer B, and output layer C according to the equations (1) to (4).

At this time, the output value of the neuron of the output layer C is compared with the teacher data d. If the difference is within an allowable value (0.001 is used in this example), the learning is terminated, and the weight at this time is The coefficient and the threshold value are stored in a storage unit of the host computer 1, that is, a memory. However, if it is larger than the allowable value, a new weight coefficient or threshold value is calculated again using the above-described equations (5) to (8), and again, the weight coefficient and the threshold value are used.
Formulas (1) to (4) are calculated, and the calculation is repeated until the difference between the output of the neuron of the output layer C and the teacher data d falls within the allowable value, and the weight coefficient Wi and the threshold value θ are determined. (S3).

Next, the host computer 1 calculates the target casting temperature in the continuous casting process, the target casting temperature in the continuous casting process, the molten steel pot from the vacuum degassing process to the continuous casting process based on the scheduled operation of the vacuum degassing process and the continuous casting process of the target charge. Time to move, the expected time to move, the expected number of continuous uses of the tundish used when casting the charge in the continuous casting process, the expected number of times the molten steel pot to be used this time, The actual time of the last hot pot in the ladle, which is the time when the molten steel was present, and the actual hot pot time, which is the idle time of the molten steel from the last tapping of the molten steel to the current receiving, are collected (S4).

Then, the new weighting coefficient Wi and the threshold value θ which have been learned from the above data are input to the input layer A of the neural network in which the data are stored. Thereby, each neuron of the input layer A, the intermediate layer B and the output layer C obtains the molten steel temperature in the vacuum degassing step according to the above equations (1) to (4), and uses the obtained molten steel temperature as the neural network. It is output as the output value Y of the output layer C (S5). This output value Y is output to the vacuum degassing process controller 3, and the vacuum degassing process controller 3 adjusts the molten steel temperature in the known vacuum degassing process such as the vacuum degassing process time and the amount of the heating material and the coolant used. Instruct.

Table 1 shows the results when the temperature in the vacuum degassing step was set in this way. As can be seen, according to this example, the hit rate within ± 5 ° C. reached 98.4%,
The hit rate was greatly improved as compared with the temperature setting by a conventional skilled operator (about 80% hit rate).

[0045]

[Table 1]

[0046]

Embodiment 2 Next, the case of setting the temperature in the converter step based on the temperature required in the vacuum degassing step will be described. First, as shown in FIG. 8, the upper computer 1 includes an input layer A having six neurons, an intermediate layer B having 50 neurons, and an output layer C having one neuron. A three-layer neural network is configured (S1).

Next, the host computer 1 calculates the tapping temperature in the converter step, the amount of alloy input in the converter step, the number of times the molten steel pot has been used, and the last time the molten steel was present in the molten steel pot during the previous use. Hot water time, empty pot time, which is the empty time of the molten steel pot from the last tapping of the molten steel until receiving it, matching, which is the time required to transport the molten steel pot from the converter process to the vacuum degassing process The time and the actual value of the molten steel temperature in the vacuum degassing process are collected and accumulated (S2).

Then, the above-mentioned collected and accumulated result data is input to the neurons of the input layer A of the neural network, and the result data is input as the teacher data d, and learning is performed in accordance with the learning method described above.
This learning is performed in exactly the same manner as the calculation of the molten steel temperature in the above-described vacuum degassing step, using equations (1) to (8).

Further, based on the scheduled operation in the converter process and the vacuum degassing process of the charge whose temperature is to be predicted, the target molten steel temperature in the vacuum degassing process, the expected alloy input amount in the converter process, and the vacuum degassing from the converter process. Scheduled matching time, which is the time required for the ladle to move to the gas process, the number of actual uses of the ladle to be used this time, the actual time of the last ladle in the ladle, which was the time the molten steel was present in the ladle during the previous use, And the actual time of the ladle, which is the empty time of the ladle from the last discharge of molten steel to the current steel reception, is collected, and the collected data is input from each neuron of the input layer A of the learned neural network. I do.

When the above data is input to the neural network, the set temperature in the converter step is calculated in accordance with the equations (1) to (4) in exactly the same manner as the temperature setting in the vacuum degassing step, and the output layer is calculated. Output from C neuron. Then, the output set temperature in the converter process is input from the host computer 1 to the converter processor 2, and the converter processor 2 is accordingly inputted.
Indicates a known converter temperature control of the converter such as the blowing time in the converter and the amounts of the heating material and the coolant.

Table 2 shows the results of setting the molten steel temperature in the converter step as in this example. As can be seen, according to this example, the hit rate within ± 10 ° C. reached 96.9%. The hit rate was improved as compared with a skilled operator (approximate hit rate of about 90%).

[0052]

[Table 2]

[0053]

According to the present invention, the measurement is easy, and the measured value is calculated using only the reliable factor data of the molten steel temperature drop. The temperature can be adjusted accurately without using a large-capacity computer, and the tapping temperature of the converter can be reduced, so that the cost of refractories such as bricks, which is the main cost of the steelmaking process, can be significantly reduced. In addition, there is a great effect that a skilled worker is not required.

[Brief description of the drawings]

FIG. 1 is a flowchart showing an outline of a method for setting a molten steel temperature according to the present invention.

FIG. 2 is a graph showing transition of molten steel temperature accompanying movement of a molten steel pot.

FIG. 3 is a block diagram showing a configuration of a neural network used in the present invention.

FIG. 4 is a block diagram showing input, output and conversion functions of the neural network of FIG. 3;

FIG. 5 is a graph showing characteristics of a conversion function used in the neuron of FIG. 4;

FIG. 6 is a block diagram illustrating a configuration example of a molten steel temperature setting system that implements the present invention.

FIG. 7 is a block diagram showing a configuration of a neural network used for setting molten steel temperature in a vacuum degassing step according to one embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a neural network used for setting a molten steel temperature in a converter process according to another embodiment of the present invention.

[Explanation of symbols]

 1: Top computer 2: Converter converter 3: Vacuum degassing converter 4: Continuous casting converter

──────────────────────────────────────────────────の Continuing from the front page (72) Inventor Atsushi Hirano 1 Nishinosu, Oita, Nippon Steel Corporation Oita Works, Ltd. (56) References JP-A-5-5121 (JP, A) Kaihei 5-195035 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) C21C 5 / 46,7 / 00 G06F 15/18 B22D 11/18 JICST file (JOIS)

Claims (1)

(57) [Claims] 1. A steelmaking process in which molten steel is received in a molten steel ladle from a converter process, and the molten steel is transferred directly or through a vacuum degassing process to a continuous casting process for processing. When the process is a post-process, when the vacuum degassing process is a self-process and the continuous casting process is a post-process, and when the converter process is a self-process and the continuous casting process is a post-process, the From the number of times the ladle has received molten steel of a certain charge in the past, the time the molten steel has been in the ladle before receiving the molten steel, and from the discharge of the previously received molten steel to the reception of the molten steel of that charge The time required to transport the molten steel pot from the self-process to the post-process, and the continuous use of the tundish which receives the molten steel of the charge when the post-process is a continuous casting process. Number of times, post-process In the case of the vacuum degassing process, the actual data of the temperature drop factor of the molten steel such as the amount of alloy injected into the molten steel in the converter process and the data such as the temperature of the molten steel when the molten steel arrives in the subsequent process are converted into a hierarchical neural network. And input the actual data of the molten steel temperature in the self-process of the charge, and obtain the weighting factor and the threshold value in the hierarchical neural network storing the respective actual data. In the input layer of the neural network based on the coefficient and the threshold value, the target temperature in the post-process of molten steel of the charge to be temperature-adjusted and the expected number of use of the molten steel pot to be used to receive the molten steel of the target charge, The time when the molten steel received last time was present, the estimated time of the molten steel pot to receive the molten steel of the target charge, and the transportation schedule from the self-process to the subsequent process In addition, if the post-process is a continuous casting process, the number of continuous use of the tundish is to be used continuously. If the post-process is a vacuum degassing process, the temperature drop of the molten steel such as the amount of alloy to be added to the molten steel in the converter process Enter the factor data, find the molten steel temperature of the target charge in the self process,
A method for adjusting the temperature of molten steel in the steelmaking process, wherein the molten steel of the target charge is adjusted to the determined molten steel temperature by the temperature adjusting means in the own process.