JPS63213609A - Method for predicting furnace heat in blast furnace - Google Patents

Abstract

PURPOSE:To accurately predict the variation of furnace heat in a blast furnace without any delaying, by adopting an ARMA model using solution loss carbon quantity or nitrogen content in top furnace gas component as a variable to an MA term. CONSTITUTION:The average value of the solution loss carbon quantity at each interval during a time by going back in a past time from the present time, is calculated and introduced as the variable to the MA term. Next, the predicting value of the molten iron temp. at each iron tapping hole at a time in the future, is found, to compare it with the control temp. In case some abnormality is recognized, warning is given and the above process is gone back to the original step, and in case any abnormality is not recognized, the predicting value of the molten iron temp. is outputted and also the ARMA model, which is gone back to the original step, is adopted. By this method, the effect of deviation of the molten iron temps. by difference of the iron tapping holes can be removed and the prediction accuracy of furnace heat in the blast furnace can be improved. Further, even in case the average value of the nitrogen content in the top furnace gas detected by a gas chromatography is used as the variable to the MA term, the same effect can be expected.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a blast furnace furnace heat prediction method for stable operation of a blast furnace.

(Prior art and its problems) It has been known for a long time that in order to maintain stable operation of a blast furnace, it is necessary to keep the temperature of hot metal constant. For this reason, blast furnace operators have always needed to predict changes in blast furnace heat.

Prediction of temperature drop is extremely important in blast furnace furnace thermal changes, especially since hot metal may solidify due to temperature drop and may no longer flow out of the blast furnace.

A method for predicting blast furnace heat is to consider the blast furnace as a black box and obtain a model through statistical analysis.
There is one based on the AR (Auto-Rearessive) model.

Although the AR model described above has excellent characteristics as a model for predicting temporal fluctuations in the actual furnace heat level of a blast furnace, it has the following problems.

FIG. 7 is a diagram showing changes over time in actual values and predicted values of hot metal temperature, in which fl1 is a curve based on the actual value and 12 is a curve based on the predicted value. As shown in the figure, the change in the predicted value lags behind the change in the actual value by approximately τ sampling time (i.e., τ × Δ is time, where the sampling interval is Δt);
There is also a tendency for the amplitude to be small. For this reason, there was a problem in that the prediction function could not be fully fulfilled, and the accuracy of the predicted value was also not sufficiently accurate.

In order to solve this problem, ARM A (Auto-
Realistic Moving Average
) (autoregressive moving average) model is published in JP-A-60-248804°T Tetsu-to-Hagane J 1984, No. 1,
Although it is disclosed on page 54, etc., the current situation is that no definitive variable to be introduced into the MA term has been found yet.

(Objective of the Invention) The object of the present invention is to provide a method for predicting blast furnace heat that can accurately predict changes in hot metal temperature without delay in prediction by solving the problems of the prior art described above. be.

(Means for Achieving the Object) In order to achieve the above object, the blast furnace furnace heat prediction method according to the present invention uses the blast furnace operation results to predict temporal fluctuations in the blast furnace furnace heat level using the ARMA model. The amount of solution loss carbon or nitrogen in the top gas component is introduced as a variable into the MA term.

(Example) It has been conventionally known that an increase or decrease in solution loss carbon m (hereinafter referred to as "Tru loss Cff1J"), which indicates the quality of the reduction state of a blast furnace, has a strong correlation with blast furnace heat. In other words, the increase in truss loss Cff1 indicates that the so-called tsuru loss reaction shown below is promoted, C+C02 → 2GO Since this reaction is an endothermic reaction, the furnace heat decreases sooner or later (in fact, after about a few hours). I will do it.

Therefore, it can be said that increases and decreases in the amount of truss loss C can be predicted with respect to changes in blast furnace heat.

Therefore, AR which introduces the average value of Tsururos 0 into the MA term
An MA model can be considered.

FIGS. 1(a) and 1(b) show changes over time between the instantaneous value of the hot metal temperature and the average value of the amount of trunnion C, respectively. The same figure (a
), Ll is the actual value and L2 is the predicted value. The amount of vine loss C is calculated based on the proportions of CO, CO2, N2, etc. in the top gas components, ventilation conditions, raw material charging conditions, etc., and the average value of vine loss C■ shown in Fig. 1(b) u (t ) is obtained by calculating the instantaneous value of Tsuruloth 0 (Ky/1-p) at each sampling time of gas chromatography, for example, and averaging this at each time width Δt1. On the other hand, the instantaneous hot metal temperature flu(t) is the sampling time Δt2
It is required for each. Note that t is the current time.

Using such variables, the predicted hot metal temperature value V(t+Δt2) at one point in time can be estimated using the following equation using the ARMA model.

V(t+Δt2) ...(1) The first term on the right side is the AR term, and the second term is the MA term.

FIG. 2 is a flowchart showing a prediction processing procedure based on equation (1). First, in step $1, go backwards from the current time t to the past period NΔt1,
The average value u(Δt1 to t−) of the crane loss Cff1 for each
k=1 to N) is calculated by N11lil. Note that this process is performed every sampling time Δt2. Next step S2
, the hot metal temperature actual value y(
t), by comparing this with the past prediction result V(t), the coefficient aj of equation (1) is sequentially determined every Δt2.

Then, in step S3, the predicted value V(t+Δt2) of the hot metal temperature one point in time is obtained from equation (1) using the coefficients a・, b・ obtained as described above, and in the next step S4, this predicted value V (t+Δt2) and the control temperature 11 are compared, and if tcH≧VLTCH (t+Δt)≧tCL, it is assumed that there is no abnormality and the process returns to step S1 to continue prediction. On the other hand, in step S4, V(t+Δ1)<1° or V(t+Δ1)
1) If >1c, it is assumed that a decrease or increase in the furnace heat will occur sooner or later, and a predicted value V (t+Δt2) is output to warn of the decrease or increase in the furnace heat. Thereafter, the process returns to step S1 and the furnace heat change is predicted while repeating steps 81 to S5.

In Figure 3, Δt1 = 30 (min), N = 2
This is a graph showing actual prediction results. In the same figure, Ll is the actual hot metal temperature value×, L2 is the predicted hot metal temperature value・
tCL'taft is the control temperature. As is clear from the figure, the control temperature t.

At the time when the actual value shows a value higher than L, it is possible to almost predict the decrease in furnace heat at one point in time, and it becomes possible to output an alarm (warning) of the decrease in furnace heat at times t1 to t4. . In addition, during this time, the control temperature t. No furnace heat rise exceeding 11 occurred.

By the way, since a blast furnace normally taps pig iron alternately through two or more tap holes, there is a deviation in the hot metal temperature at different tap holes, as shown in the change in hot metal temperature over time for each tap in Figure 4. In many cases, prediction accuracy is affected. Therefore, a method of predicting the hot metal temperature by setting up an ARMA model for each of tapholes A and B can be considered. If the tapping at one point in time is through taphole A, using the temperature yA (t) of the hot metal tapped from taphole △, yA (t+
Δt3), the hot metal temperature r! Predict 3yA, taphole B
In the case of tapping, the temperature y of hot metal tapped from taphole B
8 (t), predict the hot metal temperature y8 at one point in time using y, , (t+Δt3). In addition, aA
j, bAi, a8j, b81 are coefficients, and Δt3
is the sampling time of the hot metal temperature.

FIG. 5 is a flowchart showing the processing procedure for prediction using the ARMA model for each tapping date. In the same figure, step 3
11, going backwards from the current time t to the past by a period NΔt1,
Average value u of crane loss l for each interval Δt1 (Δt1 in t-
) (k=1 to N).

Next, in step 812, the taphole that is one point ahead is distinguished, and if the taphole that is one point ahead is taphole A, then in steps 813 and 814, the hot metal of taphole A is changed based on equation (2). Calculate the predicted temperature value V (t+Δt3) and proceed to step S15.
If the control temperature 11 and cL'cl are compared, and if V(1Δ13)<1° or yA8(1Δ1)>1oH, V(t+Δt3) is output in step S16, and the furnace heat decreases or A warning is given for the rise, and the process returns to step S11. If t≧y (t+Δt3)≧cllA tcL in step 315, it is assumed that there is no abnormality, and the process returns to step S11.

On the other hand, in step 812, if the tap point one point ahead is tap hole B, step S17. In step S18, the predicted value V8 (t+Δt3) of the hot metal temperature at the taphole B is determined based on equation (3), and in the subsequent step S19. In S20, this y (t+Δ
Based on step t3), processing similar to steps 315 and 816 described above is performed.

By providing an ARMA model for each tap date in this manner, the influence of hot metal temperature deviation due to the difference in the tap hole can be removed, and the prediction accuracy is considerably improved.

Furthermore, when calculating the average of the trail loss Cff1 for each section Δt1, the instantaneous value of the trail loss Cff1 becomes an abnormal value E1 due to noise or the like as shown in FIG. 6(a). E2 may occur. Here, the truss loss Cff1 at time j is expressed as
・, 1 sampling time Δ is the previous true loss Cff1 as X
・ Then, the absolute value Δx of the difference value of the true loss Cff1
j is ΔX, ==lX, -x-1=(4) J J J-1. By comparing this ΔX· with the threshold value ε2 as shown in FIG. 2(b), the abnormal value E1. I gave birth to E2, and the same figure (C
), a method of smoothing can be considered by replacing - with the previous value.

By applying this method, more accurate truss C
The average value of the barriers is determined, and as a result, more accurate predictions are possible.

In addition, the nitrogen base (%) in the furnace top gas detected by gas chromatography (hereinafter referred to as [gas chromatography N2ff1J) has a strong negative correlation with truss loss Cff1. A similar effect can be expected by using it in the term.

In this embodiment, the average value of the trail loss Cff1 for each section Δt1 is used as the MA term, but an instantaneous value of the trail loss Cm can also be used instead. However, since it has the disadvantage that noise components are likely to be included, the average value is more desirable.

(Effects of the Invention) As explained above, according to the present invention, an ARM using the amount of turret loss C or gas chromatography N2m as a variable in the MA term.
The A model allows for accurate prediction of blast furnace heat changes without delay.

[Brief explanation of the drawing]

1(a) and 1(b) are graphs showing changes over time in the instantaneous value of hot metal temperature and the average value of trunnion loss Cff1, respectively; FIG. Figure 3 is a graph showing actual prediction results, Figure 4 is a graph showing changes in hot metal temperature over time when the temperature of hot metal varies depending on the tap hole, and Figure 5 is another embodiment of the present invention. Flowchart showing the processing procedure of the prediction method using the ARMA method by tapping date, FIG. 6(a)
, (b), (respectively, the true loss Cf including abnormal values
A graph showing the instantaneous value of f1, the absolute value of the difference value of the amount of trailing loss C, and the instantaneous value of trailing loss Cff1 after removing abnormal values,
FIG. 7 is a graph showing prediction results based on actual hot metal temperature values based on the conventional AR method.

Claims (3)

[Claims] (1) When predicting temporal fluctuations in the blast furnace heat level using the ARMA model using blast furnace operation results, it is important to note that the amount of solution loss carbon or the amount of nitrogen in the top gas component is introduced as a variable in the MA term. Features of blast furnace furnace heat prediction method. (2) The blast furnace furnace heat prediction method according to claim 1, wherein the average value of the solution loss carbon amount or the nitrogen amount in the furnace top gas component for each predetermined period is introduced into the MA term as a variable. (3) The blast furnace furnace heat prediction method according to claim 1 or 2, wherein the ARMA model is different depending on the tap hole.