JP2000096114A - Method for predicting furnace condition in blast furnace - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a predicting method of the furnace condition in a blast furnace, by which the abnormal condition in the blast furnace can suitablly be predicted. SOLUTION: In the blast furnace, in which the furnace top pressure and the blasting flow rate are controlled to the constant, respectively, the blasting pressure and the pressure in the furnace are detected and these data with time are analyzed with a time frequency. Then, (1) an indexing is executed by emphasizing that the high frequency component in the vibration becomes higher in comparison with the stable time of the furnace condition, (2) an indexing is executed by emphasizing the imbalance degree in the ranges of the lower frequency and the higher frequency than a feature frequency generated at the stable operation time as the center or (3) an indexing is executed by recognising that the intensity of the vibration is wholly changed in comparison with the stable operation time, and these indexes are used the feature values and the abnormal condition is beforehand prediced based on the feature values.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for predicting a furnace condition in a blast furnace in which a furnace top pressure and a blown air flow rate are controlled to be constant, and more particularly, to a method for predicting an abnormal furnace condition based on time-series data of a blown pressure or a furnace pressure.

[0002]

2. Description of the Related Art Conventionally, this kind of blast furnace condition prediction method has been disclosed in Japanese Patent Publication No. 60-41123 and Japanese Patent Laid-Open No. 3-126.
806, "Iron and Steel 72 [10] (1986),
Development of Blast Furnace Abnormal Furnace Condition Prediction System P1545 ". These prediction methods manage information from many sensors installed in the furnace body in a pattern (high level, large difference in the circumferential direction, increasing tendency, etc.) that can be visually judged by an operator, and set a reference standard. The state of the furnace is determined by comparing with a value or a theoretical value, and the quality of the furnace condition is determined by comprehensively evaluating the determination results of all the sensors.

[0003] Japanese Unexamined Patent Publication No. 10-60510 also proposes a "method for predicting an abnormal blast furnace condition". This predicting method is based on a time-series signal which cannot be determined only by visually monitoring the time transition of a sensor signal. Is evaluated. Specifically, the second moment (a kind of variation) of the power of the specific frequency is evaluated.

[0004]

SUMMARY OF THE INVENTION The above mentioned Japanese Patent Publication No. Sho 60-4
No. 1123, Japanese Unexamined Patent Application Publication No. 3-126806, and "Development of a system for predicting abnormal furnace conditions in blast furnaces P1545" in "Iron and Steel 72 [10] (1986)" (A high level, a difference in the circumferential direction is enlarged, a rising trend, etc.), and it is a systematization, and a subtle change (change when observed in the time domain) existing in a time-series data sequence that cannot be judged visually ) Is not dealt with. For this reason, there is a problem that the detection of the abnormal furnace condition is limited to a case where the pattern is largely changed to such an extent that a human can judge.

[0005] The prediction method disclosed in Japanese Patent Application Laid-Open No. 10-60510 solves the above-mentioned problems. However, since the furnace condition is determined based on the following equation, the prediction method shown in FIG. However, there is a problem that it cannot cope with a change in the frequency distribution as shown in FIG. That is, this prediction method evaluates only the variation of the normalized frequency distribution, and the variation of the frequency distribution does not change. However, as shown in FIG. 5, when the shape and the amplitude of the frequency distribution change, Decreases the sensitivity.

[0006]

(Equation 1)

One of the characteristics when the furnace condition is abnormal is that the frequency components other than the specific frequency f ₀ appearing when the operation is stable increase.
It is a phenomenon that the frequency distribution decreases (the shape of the frequency distribution changes), and characteristic patterns thereof are shown in FIGS. 5A and 5B as a time series transition diagram and a frequency distribution diagram. FIG.
FIG. 5A shows an example in which the high-frequency component of the vibration becomes stronger than that in the case where the furnace condition is stabilized, and FIG. This is an example where the balance between the region and the region of lower frequency than that has changed.

Another characteristic when the furnace condition is abnormal is that the amplitude (power) of the frequency distribution changes. The characteristic pattern is shown in FIG. 5C as a time series transition diagram and a frequency distribution diagram. . FIG. 5 (C) shows an example in which the vibration intensity repeats an increase / decrease as a whole as compared with the time when the reactor condition is stable.

The present invention has been made in view of such a situation, and an object of the present invention is to provide a blast furnace furnace condition prediction method which can appropriately predict an abnormal furnace condition of a blast furnace.

[0010]

A method for predicting a blast furnace condition according to the present invention detects a blowing pressure or a pressure in a blast furnace in which a furnace top pressure and a blowing flow rate are controlled to be constant, respectively.
The time-series data is subjected to time-frequency analysis, and the shape or amplitude of the frequency distribution is determined from the analysis result, which is used as a characteristic amount, and an abnormal reactor condition is predicted in advance based on the characteristic amount.

In the present invention, for example, feature values 1 to 3 are calculated from the result of the frequency analysis as follows, and the feature values are compared with a reference value for determination. The symbols used in the following equations are defined as follows. Fi (t): feature amount P (t, f): power Dt (t, f): time derivative Dt (t, f) = P (t, f) -P (t-1, f) (2) Df (t, f): frequency derivative Df (t, f) = P (t, f) -P (t, f-1) (3) t: time f: frequency

(1) Feature 1 (F1 (t)) As shown in FIG. 1A, this feature 1 indicates that the high-frequency component of the vibration has become stronger as compared to when the reactor condition is stable. It is an exponential value with emphasis, and is obtained by calculating the sum of the amount of decrease in the component in the low frequency region and the amount of increase in the component in the high frequency region.

[0013]

(Equation 2)

In the above equation, the sum of the areas A1 and A2 of the difference between the solid line (state in a stable state) and the dashed line (state in an abnormal state) in FIG. Here, not only the increase amount (A1) of the component in the high frequency region but also the sum of the decrease amount (A2) of the component in the low frequency region is obtained.
This is because if the high frequency component of the vibration becomes strong, the component in the low frequency region will inevitably decrease,
By calculating the sum of (A1, A2), the increase of the high frequency component of the vibration is emphasized, and the periodic fluctuation at the time of the deterioration of the furnace condition is emphasized.

(2) Feature 2 (F2 (t)) As shown in FIG. 1B, this feature 2 is centered on a feature frequency that appears when the operation is stable, and a region of higher frequency than that is shown. This is an exponential value obtained by emphasizing the degree of imbalance with a lower frequency region, and is obtained by the following equation. The characteristic frequency that appears when the operation is stable corresponds to the raw material charging cycle, and can be accurately specified by managing the raw material charging intervals.

[0016]

(Equation 3)

In the above equation, the difference (f−f) between the characteristic frequency f ₀ and the characteristic frequency f ₀ in the higher frequency region and the lower frequency region is centered on the characteristic frequency f ₀ in FIG. f ₀ ) and the power (P (t, f)) are multiplied and integrated, but by calculating in this way, a change in the frequency component apart from the characteristic frequency f ₀ is emphasized. And emphasizes the periodic fluctuations when the reactor condition deteriorates.

(3) Feature 3 (F3 (t)) As shown in FIG. 1 (C), the feature 3 has an overall increase in vibration intensity compared to when the reactor condition is stable. This is an exponential value obtained by (decrease), and can be obtained by calculating the total value (total value of increase and decrease) of the change amount of the amplitude (power) of each frequency by the following equation.

[0019]

(Equation 4)

[0020]

FIG. 2 is an explanatory view showing the inside of a blast furnace to which the present invention is applied. As shown in the drawing, coke and ore are charged into the furnace of the blast furnace 10 in layers, and coke layers 11 and ore layers 12 are formed alternately. The raw material in the furnace is descending toward the lower part of the furnace at a constant descending speed when stable due to coke combustion and ore melting in the lower part of the furnace.
The temperature distribution in the furnace rises toward the lower part, such as about 2,000 degrees at the lower part of the furnace and several hundred degrees at the upper part of the furnace. About 1000-
In the 1100 degree region, the ore begins to melt, and there is a molten zone 13 in which the ventilation resistance is several hundred times that of the coke layer. The thickness distribution of the coke layer 11 and the ore layer 12 and the melting zone 13
, Especially the number of slits 14 in the coke layer of the molten zone 13 greatly affects the ventilation resistance in the furnace. Since the blast furnace performs constant control of the furnace top pressure and the constant flow rate of the blast, the change in the ventilation resistance in the furnace can be managed by the output of the blast pressure sensor 15 and the output of the shaft pressure sensor 16, that is, the blast pressure and the shaft pressure.

FIG. 3 is a block diagram showing a configuration of a system to which a method for predicting a blast furnace condition according to an embodiment of the present invention is applied. As described above, it is assumed that the blast furnace 10 is controlled so that the furnace top pressure and the flow rate of the blast are kept constant. Since these controls have been performed conventionally, the details thereof are omitted. As shown in the figure, the blast furnace 10 is provided with a blast pressure sensor 15 for detecting a blast pressure and a shaft pressure sensor 16 for detecting a shaft pressure. The data detected at is collected at regular intervals and stored. The time-frequency analysis unit 26 performs a time-frequency analysis of the data within the past fixed period from the latest value of the collected and accumulated pressure data.

Further, the frequency analysis unit 26 performs the above-mentioned (2) based on the data P (t, f) subjected to the time-frequency analysis.
Expressions (8) to (8) are calculated, and feature amounts 1 to 3 (F1 to F
3) are obtained, and their characteristic amounts 1 to 3 (F1 to F3) are obtained.
Is compared with the reference value corresponding to each of them to predict the abnormal reactor condition in advance. The feature values 1 to 3 (F1 to F3) and the result of the prediction are displayed on the monitor monitor 27.

The frequency analysis unit 26 performs the above-described calculation for each sensor to obtain the characteristic quantities 1 to 3, and obtains spatial information from a large amount of sensor information installed in the furnace body as in the related art. It is not necessary to determine patterns (circumferential distribution, radial distribution, vertical distribution, etc.) and monitor their changes. For example, if there is only one blowing pressure as sensor information, it is possible to foresee in advance, and the sensor will be sufficient It can also be applied to blast furnaces that are not installed (many blast furnaces overseas). Also, it is not always necessary to obtain all of the feature amounts 1 to 3, but it is sufficient to obtain at least one feature amount.

FIG. 4 is a timing chart showing the transition of the feature values 1 to 3 when the blow-by occurs and when the system is stable in the system of FIG. The blow-by occurrence data indicates data from the past 10 hours before the occurrence of the blow-by. The blow-by occurrence data shows a tendency that the feature amount fluctuates or increases several hours before compared with the stable time data, and the reference values indicating that there is a high risk of occurrence of the blow-through for all of the special quantities 1 to 3 are set. Was exceeded.

In the operation of a blast furnace, the operation action to prevent blow-by must start several hours before, so it is too late to take measures to prevent even if an alarm is issued immediately before blow-through. . Therefore, it is desirable that the blow-through is predicted several hours before the occurrence, but the blow-through itself is uneven distribution of the raw material in the blast furnace several hours before it occurs, occurrence of shelf hanging, unevenness of raw material unloading Since it occurs due to such factors, it is considered that prior prediction is possible in principle. Therefore, the feature amount 1
3 can be determined to have captured such an abnormal state in the furnace in advance. As described above, advance prediction of the blow-by is possible by the feature values 1 to 3, and the operator monitors these feature values 1 to 3 and avoids blow-by when there is a high risk of occurrence of the blow-by. It is now possible to take operational actions.

[0026]

As described above, according to the present invention, the blowing pressure or the pressure in the furnace is detected, the time-series data is subjected to time-frequency analysis, and the abnormal furnace condition is determined based on the periodic fluctuations appearing in the analysis result. Because the prediction is made in advance, it is possible to detect even a subtle change before the abnormal furnace condition existing in the time-series data sequence, which cannot be determined by a method such as visual monitoring of a transition diagram normally performed by an operator. In addition, since the fluctuations in the shape or amplitude of the frequency distribution that could not be detected conventionally were detected and the prior prediction was performed based on the fluctuations, the phenomenon before the abnormal reactor condition occurred was appropriately detected, The accuracy of advance prediction is improved.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram for obtaining a feature amount in the present invention.

FIG. 2 is an explanatory view showing the inside of a blast furnace to which the present invention is applied.

FIG. 3 is a block diagram showing a configuration of a system to which a method for predicting a blast furnace condition of a blast furnace according to an embodiment of the present invention is applied.

FIG. 4 is a diagram showing a transition of a feature amount obtained by the system of FIG. 3;

FIG. 5 is an explanatory view of an abnormal furnace condition of a blast furnace.

Claims (4)

[Claims] In a blast furnace in which a furnace top pressure and a blowing flow rate are controlled to be respectively constant, a blowing pressure or a pressure in the furnace is detected, time-series data thereof is subjected to time-frequency analysis, and a shape or amplitude of a frequency distribution is obtained from the analysis result. A blast furnace condition prediction method, wherein a variation of the blast furnace condition is obtained, and the variation is used as a feature value, and an abnormal furnace condition is predicted in advance based on the feature value. 2. The blast furnace furnace according to claim 1, wherein the characteristic amount is an index value that emphasizes that a high-frequency component of the vibration is higher than that in a time when the furnace condition is stable. Condition prediction method. 3. The characteristic amount is an index value obtained by emphasizing the degree of imbalance between a low frequency region and a high frequency region centering on a characteristic frequency appearing during stable operation. A method for predicting a blast furnace condition according to claim 1. 4. The blast furnace condition prediction according to claim 1, wherein the characteristic amount is an index value indicating that the vibration intensity has changed as a whole as compared with when the operation is stable. Method.