JP2017165999A - Operation method of blast furnace - Google Patents

Abstract

PROBLEM TO BE SOLVED: To specify an operation abnormal place of a blast furnace.SOLUTION: An operation method of a blast furnace includes: specifying an operation abnormal place of a blast furnace in which pressure sensors for detecting furnace internal pressure are installed in a plurality of places; and changing an operation condition of the blast furnace so as to remove abnormal factors for the operation abnormal place. Furthermore, the operation method for the blast furnace includes: a first step of obtaining individually a correlation coefficient between first furnace internal pressure detected with the one pressure sensor among the plurality of pressure sensors, and second furnace internal pressure detected with the all the remaining pressure sensors; and a second step of specifying an operation abnormal place based on the correlation coefficient thereof.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a method for operating a blast furnace in which an abnormal operation location of the blast furnace is specified and the blast furnace is operated so as to eliminate the cause of the abnormal operation.

  In order to operate the blast furnace stably, a plurality of sensors are installed in each direction and each level in the blast furnace, and the furnace condition is judged based on these information. However, it is often difficult to judge the furnace status by simply monitoring and managing each sensor information.

  Here, in Patent Document 1, the blast pressure or the pressure in the furnace is detected, the time series data is subjected to time frequency analysis, and the frequency distribution centering on the frequency component corresponding to the time interval of the blast furnace raw material charging is obtained, A method for predicting furnace abnormalities in advance based on frequency distribution is disclosed.

  In Patent Document 2, the blast pressure or the pressure in the furnace is detected, the time-series data is subjected to time-frequency analysis, the shape of the frequency distribution or the fluctuation of the amplitude is obtained from the analysis result, and this is used as the feature amount. A method for predicting abnormal furnace conditions in advance is disclosed.

  In Patent Document 3, the blast pressure or the pressure in the furnace is detected, the time series data thereof is analyzed by time frequency, and from the analysis result, at least one of the displacement of the frequency distribution and the power fluctuation index with respect to the reference frequency distribution, There is disclosed a method for predicting a furnace condition of a blast furnace based on a linear differential equation including a power spectrum at a specific frequency and based on the obtained linear differential equation.

  In Patent Document 4, frequency analysis is performed on at least one operation data among exhaust gas temperature, equipment temperature including the furnace body, molten metal temperature, blowing pressure, pressure on the inner wall of the furnace, and displacement of the deposition surface in the furnace. A method for specifying a peak corresponding to a cycle of a factor as a cause of malfunction is disclosed.

Japanese Patent No. 3319297 Japanese Patent No. 3521760 Japanese Patent No. 3487203 Japanese Patent No. 5277636

  However, even if the deterioration of the furnace condition can be detected at an early stage by the methods of Patent Documents 1 to 3, it has not been possible to identify the site of operation fluctuation. In the method of Patent Document 4, it is not possible to identify the site where the fluctuation occurs for the operation fluctuation that does not correspond to the cycle of the operation factor assumed in advance. Furthermore, in patent documents 1-4, about the operation fluctuation | variation whose period does not change, the generation | occurrence | production site | part of operation fluctuation | variation was not able to be specified.

In order to solve the above-mentioned problems, the operation method of the blast furnace according to the present invention is as follows: (1) identifying an abnormal operation location of the blast furnace in which a plurality of pressure sensors for detecting the pressure in the furnace are installed, A method of operating a blast furnace that changes the operating conditions of the blast furnace so as to eliminate the cause of the abnormality in the operation abnormal part,
A correlation coefficient between the first furnace pressure detected by any one of the plurality of pressure sensors and the second furnace pressure detected by all the remaining pressure sensors is individually determined. It has the 1st step to obtain | require, and the 2nd step which specifies an operation abnormal part based on the magnitude of these correlation coefficients, It is characterized by the above-mentioned.

  (2) When one correlation coefficient among the plurality of correlation coefficients obtained in the first step is smaller than a threshold value corresponding to the operation abnormality, the second furnace corresponding to the correlation coefficient The method for operating a blast furnace according to (1), wherein a location where the pressure is detected is identified as an abnormal operation location.

  (3) When two or more correlation coefficients among the plurality of correlation coefficients obtained in the first step are smaller than a threshold value corresponding to an operation abnormality, the second corresponding to these correlation coefficients The method for operating a blast furnace according to (1), wherein all locations where the pressure in the furnace is detected are identified as abnormal operation locations.

  (4) When two or more correlation coefficients among the plurality of correlation coefficients obtained in the first step are smaller than a threshold value corresponding to the operation abnormality, the second corresponding to these correlation coefficients The method for operating a blast furnace according to (1), wherein a location having the smallest correlation coefficient among all locations where the in-furnace pressure is detected is specified as an abnormal operation location.

  According to the present invention, it is possible to identify an abnormal operation location of the blast furnace using the correlation coefficient of the in-furnace pressure acquired at different detection positions. That is, in addition to the deterioration of the furnace condition, it is possible to identify an abnormal operation location, and thus it is possible to perform a more accurate operational abnormality recovery action.

It is the schematic of a blast furnace. It is a graph of the correlation coefficient of the furnace pressure which changes according to the value of n. It is a graph of the correlation coefficient of blowing pressure and furnace pressure (comparative example) It is the graph which investigated each correlation coefficient in the furnace pressure in the position used as a reference | standard, and the furnace pressure in another position (Example).

(Identification of abnormal operation location)
FIG. 1 is a schematic view of a blast furnace to which the present invention can be applied, and schematically shows a layer structure of a blast furnace raw material charged in the furnace. A black layer is an ore layer, and a white layer is a coke layer. The blast furnace 1 is a bell-less blast furnace having a swirl chute 12 at the furnace top 11, in which iron ore and coke, which are raw materials for the blast furnace, are charged in layers from the top of the furnace, and coke is burned by hot air blown up from the bottom of the furnace. The iron oxide in iron ore is reduced and dissolved by the high-temperature reducing gas generated in the process to produce pig iron. However, the present invention can also be applied to a bell-type blast furnace including a large bell and a small bell that can move up and down.

  A plurality of in-furnace pressure sensors 13 are provided on the furnace wall of the blast furnace 1. The installation position of each in-furnace pressure sensor 13 can be specified based on the direction and the installation height (hereinafter referred to as a level). In this embodiment, the in-furnace pressure sensors 13 are installed at a total of 16 locations, and each in-furnace pressure sensor is determined according to the direction composed of NW, SE, NE, SW and the level composed of B2, S1, S2, and S3. Thirteen installation positions can be defined. B2, S1, S2, and S3 are arranged in this order from the bottom of the furnace toward the top 11 of the furnace. The in-furnace pressure sensor 13 is usually installed in the blast furnace 1 at least 16 or more. A well-known configuration can be used for the furnace pressure sensor 13.

  Dozens of tuyere 14 are disposed in the lower part of the furnace in the circumferential direction of the furnace, and hot gas is blown from the tuyere 14 as a reducing gas. The blowing pressure of the gas blown from the tuyere 14 is detected by the blowing pressure sensor 15. In the lower part of the furnace, there is a region where an ore fusion layer having a large ventilation resistance softened and fused, called a cohesive zone, and a coke slit having a relatively small ventilation resistance derived from coke are mixed. The reducing gas that has been blown flows in the direction indicated by the arrow.

  In the above-described blast furnace 1, when the furnace condition is stable, it is considered that the blast furnace raw material descends at a constant speed without disturbing the layer structure and the cohesive zone shape. If the descending speed of the blast furnace raw material is stable, the charging period of the blast furnace raw material becomes constant, so that the pressure in the furnace also varies according to the charging period of the blast furnace raw material.

  Here, if there is a local disturbance in the descending speed of the blast furnace raw material for some reason, the layer structure is also locally disturbed, so the fluctuation cycle of the furnace pressure at that location changes, or the period does not change. However, some change occurs, such as a phase change.

  Therefore, in this embodiment, time series data of the furnace pressure of each direction and each level is acquired, and the furnace pressure detected by one reference furnace pressure sensor 13 and the remaining furnace pressure sensors 13 are used. For each combination with the detected furnace pressure, a data string consisting of two sets of numerical values is created, and the correlation coefficient of each of the two sets of data strings is calculated (corresponding to the first step). From the calculated correlation coefficient value, it is possible to identify the occurrence site of the operational fluctuation. One in-furnace pressure sensor 13 serving as a reference is arbitrary, and can be appropriately selected from a plurality of in-furnace pressure sensors 13.

The correlation coefficient will be briefly described. When a data string (x, y) = {x _i , y _i } composed of two sets of numerical values is given, the correlation coefficient of the two sets of data strings can be obtained by the following equation (1). However,
The correlation coefficient is an index indicating a correlation between two sets of data strings. The correlation coefficient takes a value between −1 and 1, when the value is close to 1, the two data strings have a positive correlation, and when it is close to −1, there is a negative correlation.

  For example, the furnace pressures at times t1 to tn detected at the detection positions NW-B2 at times t1 to tn, respectively, and the furnace pressures at times y1 to yn detected at the detection positions NW-S1 respectively at y1 to yn. When yn, {x1, y1}, {x2, y2}... {xn, yn} is given as the data string (x, y). By calculating the above-mentioned arithmetic mean of x and y from these data strings and substituting this into equation (1), the correlation coefficient of the in-furnace pressure at the detection position NW-B2 and the detection position NW-S1 is obtained. Desired. The detection period of the furnace pressure can be set to 1 minute, for example. For example, when the detection period of the furnace pressure is 1 minute and n = 5 times, correlation coefficients are sequentially obtained at a 5-minute period.

  Here, the correlation coefficient between the furnace pressure at the detection position NW-B2 (corresponding to the first furnace pressure) and the furnace pressure at the detection position NW-S1 (corresponding to the second furnace pressure) is If it is smaller than the threshold value corresponding to the operational abnormality, it is recognized that an operational abnormality has occurred at either the detection position NW-B2 or the detection position NW-S1. In this case, the in-furnace pressure (corresponding to the first in-furnace pressure) at the detection position NW-B2, which is the reference, and the in-furnace pressures (corresponding to the second in-furnace pressure) at all other detection positions. When the correlation coefficient is equal to or greater than the threshold value, the detection position NW-S1 can be specified as an operation abnormality point. On the other hand, the phase of the in-furnace pressure (corresponding to the first in-furnace pressure) at the detection position NW-B2, which is the reference, and the in-furnace pressure (corresponding to the second in-furnace pressure) at all other detection positions. When the number of relations is smaller than the threshold value, the reference detection position NW-B2 can be specified as an abnormal operation location.

  Here, when the blast furnace raw material is falling stably without disturbing the layer structure, the correlation coefficient of the data string composed of two sets of values always takes a value close to 1 in any combination. become. When the layer structure is locally disturbed, the correlation coefficient of a specific combination greatly deviates from 1. Therefore, by obtaining a correlation coefficient with respect to the time-series data of the in-furnace pressure and detecting a part deviating from 1, it is possible to specify the part where the operation fluctuation occurs.

  Specifically, the abnormal point of the operation is specified with high accuracy by setting the threshold value of the correlation coefficient to a value within the range of preferably 0.3 to 0.6, more preferably 0.35 to 0.5. can do.

  The smaller the number of data serving as the basis of Equation (1), that is, the value of n, the easier it is to obtain a sharp waveform, but the more susceptible to noise. Also, the larger the value of n, the less affected by noise, but a sharp waveform cannot be obtained. Therefore, there is an appropriate range for the number of data serving as the basis of the equation (1), that is, the value of n. FIG. 2 is a graph of the correlation coefficient that changes from moment to moment, with the horizontal axis representing time and the vertical axis representing the correlation coefficient. 2A to 2F correspond to n = 5 to 10, respectively, and it is assumed that an operation abnormality has occurred in the period T1. Each measurement time is 1 hour.

  Referring to the figure, the waveform changes broadly as the value of n increases. When n ≧ 8, the change in the correlation coefficient is small, and it is difficult to distinguish between cases where there is a correlation and cases where there is no correlation. Therefore, the number of data to which the formula (1) is applied, that is, the value of n is preferably set to about n = 5-8. By setting the value of n to about n = 5 to 8, it is possible to specify the fluctuation site with high accuracy while suppressing the influence of noise. However, since an appropriate value of n is considered to change depending on the detection period of the pressure in the furnace and the charging period of the blast furnace raw material from the top of the furnace, an appropriate value may be set according to the target data.

Further, the correlation coefficient does not necessarily have to be obtained by the equation (1), and Spearman's rank correlation coefficient represented by the following equation (2) can also be used.
Here, X _i and Y _i are obtained by rearranging x _i and y _i in ascending order, respectively. About n, since it is the same as that of Formula (1), detailed description is abbreviate | omitted.

Further, the Kendall rank correlation coefficient of the following equation (3) can also be used as the correlation coefficient.
Here, P represents the number of pairs in which the magnitude relationship between two sets of X _i and Y _i matches when two arbitrary sets of data are extracted from n sets of data. Q represents the number of pairs in which the magnitude relationship between two sets of X _i and Y _i does not match when two arbitrary sets of data are extracted from n sets of data. Since X _i , Y _i , and n have been described above, a detailed description thereof will be omitted.

  Here, a method other than the above may be used as the method for specifying the operation abnormality location based on the correlation coefficient. For example, when two or more combinations of detection positions with a correlation coefficient smaller than the threshold value occur, only the detection position with the smallest correlation coefficient may be identified as an operation abnormality location, or smaller than the threshold value. You may identify all the detection positions as an operation abnormal part. Which of these is preferable depends on the presence or absence of an action capable of simultaneously eliminating abnormalities at a plurality of locations as described below.

(Abnormality resolution action)
The fluctuation of the pressure in the furnace is caused by various causes. For example, circumferential deviation of furnace top charge distribution, furnace wall deposits, defect of furnace wall bricks, circumferential deviation of PC blowing rate per tuyere, circumferential deviation of blowing rate per tuyere, etc. And a wide range. In adopting the action for eliminating the operation abnormality, an appropriate action (corresponding to the third step) is adopted as appropriate in consideration of the operation transition. At this time, it is preferable to consider not only the fluctuation in the furnace pressure but also the change in the furnace wall temperature. As described above, when two or more combinations of detection positions having a correlation coefficient smaller than the threshold value occur, it is preferable to preferentially select an action that can simultaneously resolve abnormalities at a plurality of locations. If the appropriate action is not found, it is preferable to adopt the corresponding action in order from the detection position with the smallest correlation coefficient.

An Example is shown and this invention is demonstrated more concretely.
(Example)
We applied this method to blast furnace capacity 3700 m ³ as the target, and identify the operating abnormal location. The blast furnace is provided with a total of 16 in-furnace pressure sensors in four directions (NW, SE, NE, SW) and four levels (B2, S1, S2, S3). FIG. 3 shows the result of obtaining the correlation coefficient for the combination of the blowing pressure and the pressure in each furnace for the period in which the operation fluctuation is confirmed as a comparative example. In addition, FIG. 4 shows an example of the result of obtaining the correlation coefficient for the combination of the furnace pressure of S2-NW (S2 level, NW orientation) and other furnace pressures.

  The correlation coefficient between the blast pressure and the furnace pressure is greatly different from 1 in the charging period of about 11 minutes, but all the correlation coefficients are moving in the same way, It was difficult to identify. That is, although the pressure in the furnace fluctuates approximately with the charging period, the blowing pressure fluctuates regardless of the charging period, so it is considered that the abnormal operation location could not be identified.

  Regarding the combination of S2-NW furnace pressure and other furnace pressures, the correlation between the S1-S3 level furnace pressure is very strong and the correlation with the B2 level furnace pressure is very low. It was. In particular, the correlation between the S2-NW furnace pressure and the B2-NW furnace pressure is low. From this, it was possible to specify that the vicinity of the B2 level NW azimuth was an abnormal operation location.

  Therefore, we performed an action to eliminate the abnormality to reduce the deviation in the circumferential direction of the charge, adjusting the switching timing of the forward / reverse rotation of the furnace top turning chute and the turning start position. As a result, the correlation between the S2-NW furnace pressure and the B2-NW furnace pressure became stronger, and the operation abnormality disappeared, and stable blast furnace operation could be performed.

DESCRIPTION OF SYMBOLS 1 Blast furnace 11 Furnace top part 12 Turning chute 13 Furnace pressure sensor 14 tuyere 15 Blower pressure sensor

Claims (4)

Blast furnace operation that identifies the abnormal operation location of the blast furnace where multiple pressure sensors that detect the pressure in the furnace are installed, and changes the operating conditions of the blast furnace so as to eliminate the cause of the abnormal operation of the specified abnormal operation location A method,
A correlation coefficient between the first furnace pressure detected by any one of the plurality of pressure sensors and the second furnace pressure detected by all the remaining pressure sensors is individually determined. A first step to find,
On the basis of the magnitude of these correlation coefficients, a second step of identifying an abnormal operation location;
A method of operating a blast furnace, comprising:   When one correlation coefficient of the plurality of correlation coefficients obtained in the first step is smaller than a threshold value corresponding to the operation abnormality, the second furnace pressure corresponding to the correlation coefficient is detected. 2. The method of operating a blast furnace according to claim 1, wherein the specified location is identified as an abnormal operation location.   In a case where two or more correlation coefficients among the plurality of correlation coefficients obtained in the first step are smaller than a threshold value corresponding to an operation abnormality, the second furnace corresponding to these correlation coefficients is used. 2. The method for operating a blast furnace according to claim 1, wherein all locations where pressure is detected are identified as abnormal operation locations. In a case where two or more correlation coefficients among the plurality of correlation coefficients obtained in the first step are smaller than a threshold value corresponding to an operation abnormality, the second furnace corresponding to these correlation coefficients is used. The method for operating a blast furnace according to claim 1, characterized in that, among all locations where pressure is detected, a location having the smallest correlation coefficient is identified as an abnormal operation location.