JP6690081B2 - Operation status evaluation system - Google Patents

Description

  TECHNICAL FIELD The present invention relates to an operation status evaluation system that predicts the scale and risk of “blown-out” that occurs in a blast furnace that produces pig iron, and notifies the operator of the details of the prediction.

  Conventionally, in a blast furnace, iron ore, coke, furnace interior contents such as limestone are charged in layers from the upper part, hot air is blown from the lower part, and a series of reactions such as reduction or melting of the iron ore is performed, Manufactures pig iron. Regarding the furnace interior charge, the charging position and particle size are adjusted so that an appropriate gas flow velocity distribution can be obtained in the radial direction in the blast furnace. Regarding the hot air to be blown, the blowing conditions (blast temperature, blowing moisture, etc.) from the tuyere installed on the side wall of the lower part of the blast furnace are appropriately controlled. In other words, in the blast furnace, such control is appropriately performed, and a series of reactions such as reduction and dissolution of iron ore are efficiently progressed, and operation is performed so that a desired pig iron temperature (pilot temperature) can be secured. .

By the way, during the operation of the blast furnace, an event called "blown through" may occur. The blow-through is a phenomenon in which a high-temperature gas is discharged from the top of a furnace by unevenly distributing gas in a relatively well-ventilated portion in a blast furnace with poor air permeability.
If a blow-through occurs, heat exchange between the solid and the gas will not be sufficiently carried out, and the gas will go directly to the upper part of the blast furnace at a high temperature, causing a drop in pig iron temperature, etc. I will not be missed. Therefore, it is very important to predict the possibility of blow-by from the viewpoint of stable operation of the blast furnace.

For the above reasons, the prediction technology for blow-through in blast furnace operation has already been developed.
For example, Patent Document 1 discloses a method of measuring a physical quantity in the blast furnace during operation of a blast furnace that produces hot metal from a charging material charged in the furnace, and predicting occurrence of blow-through based on the result. . In the method of Patent Document 1, the physical quantity is measured with time at two positions that are appropriately separated in the height direction of the blast furnace, the difference between the obtained physical quantities is calculated, and the calculated positional difference is calculated. A temporal difference, which is a difference from the positional difference at different times, is obtained, and it is determined whether or not blow-through occurs by using the obtained temporal difference and the positional difference.

JP-A-11-140520

Conventionally, in blast furnace operation, based on the experience accumulated by the operator, the operating condition is judged from various sensor information (for example, the temperature of the furnace wall), and various operating conditions are ensured so that the blast furnace operates stably. (Blower conditions, the amount of furnace interior charge, etc.) are controlled.
However, in recent years, a change in the operating environment may occur, for example, as an unexperienced raw material that has not been used conventionally (for example, iron ore or sintered ore) is used as a raw material in blast furnace operation. If the operating conditions cannot be appropriately controlled in response to such changes, there is a possibility that blast furnace trouble such as blow through may occur. When the operating operator is inexperienced, or even an experienced operating operator may have difficulty in appropriately determining the operating state and controlling the operating conditions.

Under such circumstances, especially considering the prevention of a sudden in-furnace gas release event called "blowing through", it is possible to adopt the method of predicting blowing through according to the technique of Patent Document 1 above. Conceivable.
However, even if the technique of Patent Document 1 is adopted, it seems difficult to actually prevent the blow through. The reason is that it may be possible to determine whether or not a blow-through occurs by the technique of Patent Document 1 described above, but the amount of heat required to compensate the heat released to the outside with the gas due to the blow-through, that is, " This is because it is not possible to predict the scale of the atrium.

  Therefore, in view of the above problems, the present invention not only risks the occurrence of blow-through when operating a blast furnace, but also makes a scale prediction when a blow-through occurs, and displays the prediction result to the operator. It is an object of the present invention to provide an operating condition evaluation system capable of more reliably preventing blow-throughs and troubles such as a decrease in furnace heat due to occurrence of blow-throughs.

In order to achieve the above object, the following technical measures are taken in the present invention.
An operation status evaluation system according to the present invention includes a data storage unit that stores, as time-series data, a value periodically measured by a sensor mounted on a blast furnace and a value calculated from the measured value, and the data storage. And a display unit for displaying a value stored in the data storage unit and a prediction result predicted by the prediction unit so that an operator can confirm the value. , And the prediction unit is a data acquisition unit that acquires time-series data from the blast furnace, and a fluctuation of pressure existing in the time-series data around the blast furnace and in a height direction, and a furnace wall temperature. obtains an index value that represents the ventilation state by using a preparative, furnace pressure, feed air pressure, obtains a score indicating a poor furnace situation with top gas temperature, and the staves temperature, using the score and the index value The danger of a void Possess an analysis section for calculating a degree, wherein the analysis unit has a size index value indicating the heat loss degree when blow with stave temperature in the time-series data is generated, the blast furnace When the temperature of the upper part and the temperature distribution in the height direction in the upper part of the blast furnace have abnormal values, the scale index value is increased .

Incidentally, preferably, the display unit, the risk of pre-Symbol blow is provided with a means for performing a display indicating the alarm issued when exceeding the threshold value is predetermined, calculated retroactively period index while the average value calculated exceeds the predetermined value from continues the alarm display indicating the risk of the blow, the alarm is issued, and the wind reduction when the furnace pressure exceeds a threshold value may If you instruct.

In addition, preferably, the prediction unit is large when the deviation from the normal value for each of the fluctuation of the furnace body pressure, the circumferential bias of the stave temperature, and the temperature distribution change of the stave temperature in the height direction. It is preferable that the index which is a value is determined and the minimum value of these three indexes is set as the index value that represents the poor ventilation state.
In addition, preferably, the predicting unit, based on the distribution of the blast furnace stave temperature in the height direction, that heat is estimated to be accumulated in the upper part, the position of the root of the cohesive zone in the blast furnace is usually Is assumed to be higher than normal, the blast pressure is higher than normal, and the furnace top gas temperature is higher than normal. However, it is preferable that the presence / absence of these five states is calculated as a score.

Contact name Preferably, the index value, the score, and an index combining the scale index value, may be calculated as a blow-risk indicator.

  According to the operation status evaluation system according to the present invention, when operating a blast furnace, not only the risk of occurrence of blow-through, but also the scale prediction when the blow-through occurs, and the prediction result is displayed to the operator. As a result, it is possible to more reliably prevent the blow through.

It is what showed the conceptual diagram of the operating condition evaluation system of the 1st technical example. It is the figure which showed the cross section of the blast furnace in which the operating condition evaluation system of a 1st technical example is provided. It is the figure which showed the relationship between furnace wall temperature and the frequency | count of the generation of the temperature. It is a figure showing the relation between furnace wall temperature and blow-through scale prediction score. It is a figure showing the relation between the amount of wind reduction and a blow-through scale prediction score. It is the figure which showed an example of the screen displayed on a display monitor. It is a conceptual diagram of an operation status evaluation system of a second technical example. It is the figure which showed the example of arrangement | positioning of the sensor in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for judging the danger of blow-by in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the furnace body pressure index value in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the gravity center index value in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the temperature difference index value in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the furnace upper temperature score in the operation condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the furnace middle part high temperature score and the specific direction protrusion score in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the ventilation pressure score in the operating condition evaluation system of a 2nd technical example. It is the figure which showed the processing flow for calculating the furnace top gas temperature score in the operating condition evaluation system of a 2nd technical example. It is a figure used for standardization of the index value in the operating condition evaluation system of the 2nd technical example. It is what showed the conceptual diagram of the operating condition evaluation system of the 3rd technical example. It is the figure which showed the example of arrangement | positioning of the sensor in the operating condition evaluation system of a 3rd technical example. It is the figure which showed the processing flow for instructing the wind reduction in the operation condition evaluation system of a 3rd technical example. It is the figure which showed the processing flow for calculating the furnace body pressure calculation value in the operating condition evaluation system of a 3rd technical example. It is the figure which showed the temperature difference reduction by the rise in furnace wall temperature and the center of gravity shift in the operating condition evaluation system of the 3rd technical example. It is the figure which showed the scaled cumulative frequency distribution in the operating condition evaluation system of a 3rd technical example. It is the figure which showed the score function used with the operating condition evaluation system of a 3rd technical example. It is the figure which showed the unsteady index | index used by the operating condition evaluation system of a 3rd technical example. It is the figure which showed the furnace height direction distribution pattern of the stave temperature in a 4th technical example operating condition evaluation system. It is the figure which showed the state transition in a blast furnace typically. It is the figure which showed the logic of the operating condition evaluation system of the 4th technical example which considered the state transition. It is the flowchart which showed the processing procedure of the operating condition evaluation system of the 5th technical example. It is a figure showing the relation between a blow-through score and an alarm announcement section. It is the figure which showed the example of the state transition series in the operating condition evaluation system of the 6th technical example.

An embodiment of a blast furnace operation status evaluation system according to the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic diagram showing a prediction system of the present invention and blast furnace equipment provided with the prediction system.
As shown in FIG. 1, the blast furnace facility has a blast furnace 1 (blast furnace main body) as its core.
The blast furnace 1 has a vertical cylindrical furnace body whose outside is covered with a steel skin 2 and whose inside is lined with a refractory 3 (refractory brick). The furnace body extends downward from the upper part to the lower part, and has a shaft part, a straight belly part from the top, a bosch part that expands upward, and a bottom part of the hearth at the bottom. In each case, a refractory 3 is attached to the inner surface of the iron shell 2, and a water-cooled metal object called a stave may be embedded between the iron shell 2 and the refractory 3 to allow cooling. In some cases, a cooling plate is embedded to allow cooling.

The side wall of the hearth floor is radially provided with openings (tuyere 4) for blowing hot air or pulverized coal into the furnace. At the time of operation, in the blast furnace 1 and at the tip of the tuyere 4, a blown air from the tuyere 4 forms a hollow portion called so-called "raceway 5" in which coke exists in a very sparse state. There is.
Further, the side wall of the hearth is provided with a tap hole for taking out pig iron S (hot metal). The pig iron S is taken out from this tap hole.

The equipment attached to the blast furnace 1 is a belt conveyor that conveys raw materials such as iron ore and coke to the top of the blast furnace 1, a hopper that temporarily stores the conveyed raw materials, and a proper distribution of the raw materials in the radial direction inside the furnace. There are bell-type or bell-less type charging devices for charging, and hot-air stoves that produce hot air.
Using a charging device, which is one of the auxiliary equipment, iron ore and a reducing material that also serves as a fuel such as coke are charged from the top of the blast furnace 1 in a layered manner. In addition, limestone is added for the purpose of removing impurities. Hot air is blown into the blast furnace 1 to which the raw material is supplied from the tuyere 4 to burn the coke therein, and a series of reactions such as reduction and dissolution of iron ore are performed to produce pig iron S. .

When operating the blast furnace 1 described above, the operator M operates under operating conditions so that a series of reactions such as reduction and dissolution of iron ore proceed efficiently and a desired temperature of the pig iron S (piping temperature) can be secured. Are in control.
For example, since the pig iron S is taken out from the blast furnace 1 at intervals of several hours (for example, 30 minutes to 2 hours), when performing the blast furnace operation, while aiming at the temperature of the pig iron S to be tapped next, the iron ore Control the amount of stones and coke. In addition, the temperature of hot air blown from the tuyere 4 (blast temperature), added moisture, the amount of pulverized coal blown, and the like are also controlled.

Therefore, the operation status evaluation system of the present invention performs not only the risk of occurrence of blow-through at the time of operating the blast furnace 1, but also the scale prediction when the blow-through occurs, and displays the prediction result to the operator. This makes it possible to more reliably prevent blowouts and troubles resulting from them.
Hereinafter, a first technical example of the operation status evaluation system for the blast furnace 1 of the present invention will be described.

  The present invention is based on the knowledge obtained through analysis of the operation results of the blast furnace 1, that is, "when the blow-through occurs in the blast furnace 1, the temperature distribution of the furnace wall is often far from the normal state". It is based on. Specifically, the present invention states that “the temperature of the stave buried in the furnace wall becomes extremely higher than usual regardless of the height of the blast furnace 1 in the vertical direction, in other words, when the entire furnace wall becomes high in temperature, It is based on the knowledge that the scale of blow-through that occurs in the blast furnace 1, that is, the amount of heat lost during blow-through increases, and the risk of operational trouble increases. The operation status evaluation system of the first technical example reflects such knowledge.

  Specifically, the operation status evaluation system of the first technical example is provided in a blast furnace that produces pig iron, and predicts troubles that occur in the blast furnace. The operator can confirm the display and the blast furnace. And a prediction unit that makes a prediction regarding “blown-out”, which is one of the problems. Then, the predicting unit predicts the scale of the "blown-out" when the "blown-out" occurs in the current operating condition, and calculates the possibility of the "blown-out". In other words, the display unit of the first technical example displays both the scale of the “blown-out” predicted by the prediction unit and the possibility of occurrence of the “blown-out”.

That is, the prediction unit has a value of 0 or more and 1 or less, and is supposed to calculate the “small-air-scale prediction score” indicating the scale of the “small”, and the display unit calculates “small-air” calculated by the prediction unit. It is configured to display "scale prediction score".
Further, the prediction unit is supposed to calculate a “blowing occurrence prediction score” indicating the possibility of occurrence of “blowing”, and the display unit displays the “blowing occurrence prediction score” calculated by the prediction unit. It is also a thing.

The above-mentioned “blown-out scale prediction score” is calculated based on a plurality of furnace wall temperatures along the height direction of the blast furnace. Specifically, the prediction unit, based on the difference between the average value of the plurality of furnace wall temperature along the height direction of the blast furnace, and the average value of the plurality of furnace wall temperature along the height direction of the blast furnace, Calculate the blowout scale prediction score.
In addition, the display unit corresponds to the time series data of the “air blowout scale prediction score” predicted by the prediction unit, the time series data of the “air blowout occurrence prediction score” calculated by the prediction unit, and each time of the time series data. It is configured to display "inside blast furnace information", which indicates the status inside the blast furnace. This "inside blast furnace information" is calculated as a value indicating where in the distribution of past operation results the situation in the blast furnace corresponding to each time of the time series data is. Furthermore, in calculating the “information inside the blast furnace”, the pressure inside the blast furnace and the temperature inside the blast furnace can be used as past operation results.

The operating condition evaluation system of the above-mentioned first technical example will be described in more detail.
The operation status evaluation system of the first technical example includes a display unit (prediction monitor 11) that can be confirmed by the operator M, and a prediction unit 13 (prediction model) that predicts "blowing through" that is one of the troubles in the blast furnace 1. And have.
The prediction monitor 11 includes a liquid crystal monitor and a CRT monitor and is installed in the control room. The operator M is always in a state of being able to view the prediction monitor 11.

In addition, the predicting unit 13 predicts the scale of the “air blow-out” (calculation of a prediction model) when the “air blow-through” occurs in the current operating condition, and calculates the possibility of occurrence of the “air blow-through”. Done.
The prediction model is realized in the form of software in the computer 12 or the computer provided in the prediction system 10.

Specifically, the computer 12 includes an arithmetic device and a storage device therein, and an input device for receiving an instruction to the computer 12 and a signal from the outside and an output device for displaying an arithmetic result. ing.
The prediction model is programmed and stored in a storage device including a hard disk, and the prediction model is calculated by a calculation device including an MPU or the like. In the calculation, the actual value obtained from the operating blast furnace 1 (the actual value of the plurality of furnace wall temperatures along the height direction of the blast furnace 1) is input through an input device composed of an AD conversion board or the like. Is adopted as the condition value of the prediction model. The actual value may be directly input by an input device such as a keyboard.

The result (predicted value of the prediction model) calculated by the calculation device is recorded in the storage device and displayed on the output device, that is, the prediction monitor 11. The actual value of the blast furnace 1 input through the input device may be simultaneously displayed on the prediction monitor 11.
In the case of the first technical example, as the prediction information calculated by the prediction model and displayed on the prediction monitor 11, a “blown-around scale prediction score” indicating a scale of the “blunder” with a value of 0 or more and 1 or less is adopted. ing. This blow-through scale prediction score is calculated by the prediction unit 13 based on a plurality of furnace wall temperatures along the height direction of the blast furnace 1. Specifically, the prediction model is such that the prediction unit 13 calculates an average value of a plurality of furnace wall temperatures along the height direction of the blast furnace 1 and an average value of a plurality of furnace wall temperatures along the height direction of the blast furnace 1. It is configured to calculate a blow-through scale prediction score based on the difference.

Hereinafter, the procedure for obtaining the “blown-out scale prediction score” will be described in detail.
In the prediction model, the blow-through scale prediction score is calculated through the following steps 1 to 5.
First, in step 1, the furnace wall temperature of the blast furnace 1 is measured at a plurality of points in the height direction. Specifically, a water cooling metal fitting called a stave is embedded in the furnace wall of the blast furnace 1, and the temperature of this stave is used as the furnace wall temperature.

As shown in FIG. 2, in the case of the first technical example, eight measurement points of B3, A1 to A6, and skin flow are provided from the lower position of the blast furnace 1, and among these eight measurement points, A4 to A6. In each of the above, the measured stave temperature is used.
When the blast furnace 1 is viewed in cross section, a plurality of staves exist at equal intervals in the circumferential direction. Therefore, the average value of the temperatures of the plurality of staves should be calculated.

In step 2, noise cut processing (low-pass filter processing) is performed on the calculated average value of the stave temperature.
That is, the moving average of the past 240 minutes is taken for each time. The stave temperature after moving average (in other words, furnace wall temperature) is as follows.
・ A6 stave temperature average value: StvT_A6 (t)
・ A5 average stave temperature: StvT_A5 (t)
・ A4 stave temperature average value: StvT_A4 (t)
-A6 stave temperature average value minus A4 stave temperature average value (difference between A6-A4 stave temperature average values): StvT_Diff (t)
T in the above variables is the time.

Note that, in addition to the difference in average stave temperature of A6-A4, the difference in average stave temperature of A5-A4 or the difference in average stave temperature of A6-A5 may be obtained.
Next, in step 3, as shown in FIG. 3, a histogram (frequency distribution) at each position is considered based on the average value of the measured stave temperature, and the following points in the histogram are obtained.

・ Peak position value or average value in the histogram: TOK
・ Position where cumulative frequency is 90% or more: TNG0
・ Position where cumulative frequency is less than 10%: TNG1
In this histogram, the fact that the average stave temperature value takes a peak position or an average value indicates that the blast furnace 1 is operating normally, and the average stave temperature exceeds 90% cumulative frequency, or When the cumulative frequency of taking a value smaller than the value reaches a value of less than 10%, it means that the blast furnace 1 is in a state far from the normal state, and there is a high risk that some trouble will occur. ing.

Similarly, a histogram (frequency distribution) at each position is considered based on the difference between the average values of the measured stave temperatures, and the following points in the histogram are obtained.
・ Peak position value or average value in the histogram: DTOK
・ Position where cumulative frequency is 90% or more: DTNG0
・ Position where cumulative frequency is less than 10%: DTNG1
In the histogram, the fact that the difference in average stave temperature takes the value of the peak position or the average value indicates that the blast furnace 1 is operating normally, and the difference in average stave temperature is the average value of stave temperature. The cumulative frequency exceeds 90%, or the cumulative frequency is less than 10%, which means that the blast furnace 1 is far from the normal state and some trouble occurs. It means the risk is high.

In step 4, as shown in FIG. 4, for each of the stave temperature average values StvT_A6 (t), StvT_A5 (t), StvT_A4 (t), and StvT_Diff (t) calculated in step 3, formula (1-1) ~ According to Formula (1-3) and Formula (2), the blow-through scale prediction score for each measurement point (for each height) is calculated.
Score_A4 = F ((StvT_A4 (t) -TOK) / (TNG0 -TOK)) (1-1)
Score_A5 = F ((StvT_A5 (t) -TOK) / (TNG0 -TOK)) (1-2)
Score_A6 = F ((StvT_A6 (t) -TOK) / (TNG0 -TOK)) (1-3)
Score_Diff = F ((StvT_Diff (t) -DTOK) / (DTNG1-DTOK)) (2)
However, F (X) is
If X <0, then F (X) = 0
If 0 < X < 1, then F (X) = X
If X> 1, F (X) = 1
Is a function.
In the blow-through scale prediction score calculated based on equations (1-1) to (1-3), the peak position or average value TOK in the histogram is the furnace wall temperature during normal operation, so the temperature average of the stave is When the value = TOK, the blow-through scale prediction score is 0. On the other hand, when the average temperature value of the stave becomes equal to or higher than TNG0, the blast furnace 1 is obviously not in normal operation, and the prediction score is 1 to represent the situation. The values of TOK and TNG0 are different in A4, A5, and A6 because the original data sets are different.

  Further, in the blow-through scale prediction score calculated based on equation (2), the peak position or average value DTOK in the histogram is the difference in the furnace wall temperature during normal operation, so the difference in the average temperature value of the stave is In the case of DTOK, the blowout scale prediction score is 0. On the other hand, when the difference in average stave temperature is less than or equal to DTNG 1, the blast furnace 1 is clearly not in normal operation, and the blow-through scale prediction score is 1 to indicate the situation.

In step 5, with respect to the four blow-through size prediction scores of Score_A6, Score_A5, Score_A4, and Score_Diff obtained by the formulas (1-1) to (1-3) and the formula (2), the formula (3) is used, Calculate the average Score_Mean.

Score_Mean = (Score_A4 + Score_A5 + Score_A6 + Score_Diff) / 4 (3)

The obtained average Score_Mean is the “open-air scale prediction score” displayed on the prediction monitor 11.

  By performing the above calculation, it is possible to calculate the “blown-out scale prediction score”, and the calculated blow-through scale prediction score is displayed on the prediction monitor 11. When the blow-through scale prediction score is close to 0, it is predicted that “the blow-through occurrence scale is extremely small”. On the contrary, when the blowout scale prediction score is a value close to 1, it is predicted that “when a blowout occurs, the scale is extremely large”.

By looking at the blow-through scale prediction score displayed on the prediction monitor 11, the operator M can instantly know the predicted value of the blow-through trouble scale. Note that the prediction monitor 11 may also display a "diagram showing the relationship between the wind reduction amount and the blow-through scale prediction score (Score_Mean)" as shown in FIG.
FIG. 5 is a diagram showing the relationship between the amount of wind reduction and the blow-through scale prediction score (Score_Mean).

This figure is created based on the data at the time of the blow-through that occurred in the past, and the points plotted in the figure are examples of the blow-through that occurred in the past. The horizontal axis of the figure shows the actual wind reduction situation (air reduction amount) in the case of blow-through, and the vertical axis shows the value of the blow-through scale prediction score calculated based on the data of the blow-through case. Is.
As shown in FIG. 5, the points plotted in the figure are the prediction scores of the past cases, which are calculated according to the procedure for obtaining the above-mentioned “blown-out scale prediction score”. For example, the □ points plotted in the figure are the prediction scores 40 minutes before the blowout, and the ◇ points are the prediction scores 10 minutes before the blowout.

Specifically, the past cases in which the prediction scores are plotted at 0.4 to 0.6 indicate that the blow-through in the blast furnace 1 was small. At this time, the operator M looked at the prediction monitor 11 on which the prediction score was displayed and slightly reduced the amount of hot air sent to the blast furnace 1 (small air reduction amount) to suppress blow-through in the blast furnace 1.
In addition, the past cases in which the prediction scores are plotted in the range of 0.6 to 0.8 indicate that the blow through in the blast furnace 1 was of a medium scale. At this time, the operator M looked at the prediction monitor 11 on which the prediction score was displayed, reduced the amount of hot air sent into the blast furnace 1 (during the amount of reduced air flow), and suppressed blow-through in the blast furnace 1.

Further, the past cases in which the prediction scores are plotted at 0.8 or higher indicate that the blow through in the blast furnace 1 was large. At this time, the operator M looks at the prediction monitor 11 on which the prediction score is displayed and largely reduces the amount of hot air sent to the blast furnace 1 (large amount of air reduction) to suppress blow-through in the blast furnace 1.
It should be noted that these past blowout cases (points □, points ◇) may be linearly interpolated and displayed. FIG. 5 shows a linearly interpolated prediction score (dot) 10 minutes before the blow-through (thick broken line).

Further, the amount of heat compensation at that time (for example, the amount of increase in the coke ratio) may be written together with the amount of wind reduction at the point position in the figure.
By displaying this figure on the prediction monitor 11, the predicted blow-through scale prediction score and the amount of wind reduction estimated to be necessary when the blow-through predicted by the predicted score actually occur are presented to the operator M. Will be shown against.

Next, the procedure for obtaining the “blowing occurrence prediction score” that numerically indicates the possibility of occurrence of “blowing” will be described in detail.
The “blown-out occurrence prediction score” indicates the possibility of occurrence of “blown-out” with a value of 0 or more and 1 or less, and is calculated by the following procedure.
First, a "furnace pressure fluctuation index" (variation level value) indicating pressure fluctuations in the blast furnace 1 is measured by using the measured values of the furnace pressure in the surroundings and the height direction of the blast furnace 1 and the measured values of the furnace wall temperature. calculate.

In calculating this “reactor body pressure fluctuation index”, for example, B3 stave, A2 stave, A5 stave, A6 stave, pressure values at each stave (pressure average values in four directions around the stave) Ask for. The variance of the obtained pressure average value in the blast furnace 1 in the past fixed time (past 30 minutes, past 60 minutes, past 120 minutes, past 240 minutes, etc.) is calculated.
After that, the calculated dispersion, that is, the amount of pressure fluctuation in the blast furnace 1 is standardized, and is divided into four stages from 0 point (small fluctuation) to 3 points (large fluctuation) in order from the smallest fluctuation of the pressure value.

Then, by summing the standardized pressure value fluctuations in the height direction (summing over all staves), there are 13 levels from 0 point (small fluctuation) to 12 points (large fluctuation). The value of 0 to 12 points is the “furnace body pressure fluctuation index”.
Next, the "index value of the furnace temperature distribution in the circumferential direction of the blast furnace 1" (the position of the center of gravity of the stave temperature in the circumferential direction of the blast furnace 1) is calculated.

  In calculating the "index value of the temperature distribution in the furnace in the circumferential direction of the blast furnace 1," for example, the stave of A2, the stave of A3, the stave of A4, the furnace temperature in the circumferential direction of each blast furnace 1 are measured. , Consider the radar distribution chart (radar chart) of the measured furnace temperature. The center of gravity position of the in-core temperature data in this radar distribution map is obtained, and the deviation between the obtained center of gravity position and the center position of the radar distribution map is obtained. The obtained deviation is the "index value of the furnace temperature distribution in the circumferential direction of the blast furnace 1."

Then, the "index value of the furnace temperature distribution in the height direction of the blast furnace 1" (difference in stave temperature average value in the height direction) is calculated.
In calculating the "index value of the temperature distribution in the furnace in the height direction of the blast furnace 1," for example, the A2 stave, the A4 stave, and the furnace temperature in the respective circumferential directions are measured and measured along the circumferential direction. Calculate the average furnace temperature for each stave.

Then, the difference in the average value of the furnace temperature between the different staves is calculated. For example, the difference between the average value of the in-furnace temperature of the A2 stave and the average value of the in-furnace temperature of the A4 stave is calculated. The value of this difference is defined as "index value of temperature distribution in the furnace in the height direction of the blast furnace 1."
The values of the “reactor pressure fluctuation index”, “index value of furnace temperature distribution in the circumferential direction of blast furnace 1”, and “index value of furnace temperature distribution in the height direction of blast furnace 1” obtained as described above Each is normalized to a value of 0 or more and 1 or less, and the normalized "reactor body pressure fluctuation index", "index value of furnace temperature distribution in the circumferential direction of the blast furnace 1", "furnace temperature distribution in the height direction of the blast furnace 1" The “blown-out occurrence prediction score” is calculated by selecting the maximum value of the “index value” and by obtaining the average value of each index.

When the blowout occurrence prediction score is close to 0, it is predicted that “the blowout occurrence probability is small”, and conversely, when the blowout occurrence prediction score is a value close to 1, “the blowout possibility may occur. Expected to be high. " The calculated blowout occurrence prediction score is displayed on the prediction monitor 11.
As shown in FIG. 6, the prediction monitor 11 (display) displays the scale of the “blown-out” predicted by the prediction unit 13 and the possibility of occurrence of the “blown-out”. Is the time series data of the “blowing scale prediction score” predicted by the prediction unit 13, the time series data of the “blunder occurrence prediction score” calculated by the prediction unit 13, and the blast furnace corresponding to each time of the time series data. “Inside blast furnace information” indicating the situation in 1 is displayed.

Further, the prediction monitor 11 of the present embodiment also displays information on the flow rate of hot air blown into the blast furnace 1.
As described above, the blowout occurrence prediction score indicates the possibility of occurrence of "blowing through" with a value of 0 or more and 1 or less, and when this score is less than 0.4, "the possibility of occurrence of blowing through" Is small, and if the score is 0.4 or more and less than 0.8, it is predicted that “a void may occur”, and if this score is a value of 0.8 or more and close to 1, it is “a void. Is likely to occur. ”

The time change (time series data) of the blowout occurrence prediction score is shown in the graph on the upper right side of FIG.
As shown in the upper right part of FIG. 6, the time-series data of the “blown-out occurrence prediction score” indicates a numerical change in the blow-through occurrence prediction score at each time during the operation of the blast furnace. It shows that it changes greatly depending on the changes in the operating environment.

In the time-series data of the “blown-out occurrence prediction score” on the upper right side of FIG. 6, for example, the score value rises or falls between 0 and 0.4 between about 9:00 and 17:30. I am. In other words, it can be confirmed that the normal blast furnace operation is performed from about 9:00 to about 17:30.
However, between about 17:30 and about 3:00 the next day, the score value often exists within the range of 0.4 to 0.8 (the range surrounded by light gray in FIG. 6). . That is, it can be confirmed that blow-by may occur from about 17:30 to about 3:00 the next day.

Further, between about 1 o'clock and 3 o'clock of the next day, the numerical value of the score often exists within the range of 0.8 to 1 (the range surrounded by dark gray in FIG. 6). In other words, there is a high possibility that blow-through has occurred between 1:00 and 3:00 the next day.
As described above, the blow-through scale prediction score indicates the scale of the "open-air" with a value of 0 or more and 1 or less. If this score is less than 0.6, "the size of the blow-through is extremely large. If the score is 0.6 or more and less than 0.8, the score is predicted to be "medium when the blow-through occurs," and the score is 0.8 or more and close to 1. In this case, it is predicted that “when a blow-through occurs, the scale is extremely large”.

The change over time (time-series data) of the blowout scale prediction score is shown in the form of a graph in the middle part on the right side of FIG.
As shown in the middle part on the right side of FIG. 6, the time series data of the “blown-out scale prediction score” indicates the numerical change of the blow-through scale prediction score at each time during the blast furnace operation. It shows that it changes greatly depending on the changes in the operating environment.

In the time-series data of the “open-air scale prediction score” in FIG. 6, for example, the score value rises or falls between 0 and 0.6 between about 9:00 and 17:30. There is. In other words, it can be confirmed that the normal blast furnace operation is performed from about 9:00 to about 17:30.
However, between about 17:30 and about 1:00 the next day, the score value is often within the range of 0.6 to 0.8 (the range surrounded by light gray in FIG. 6). . In other words, it is possible to confirm that when a blow-through occurs between about 17:30 and about 1:00 the following day, the scale is medium.

Further, the hot air flow rate information is also displayed on the display monitor of the present technology example.
As shown in the lower right part of FIG. 6, the hot air flow rate information indicates a change over time in the flow rate of the hot air blown into the blast furnace 1, and the hot air flow rate information indicates the blow-through scale prediction score and the blow-through occurrence prediction score. If there is a possibility that the air flow will occur, the flow rate of hot air is adjusted in order to prevent blow-by, and the information of the adjustment result (current air flow rate) is displayed.

  Further, the display monitor of the present technology example is also configured to display information about the inside of the blast furnace. Blast furnace information is displayed on the left side of the display monitor. The "inside blast furnace" shown on the left side of FIG. 6 is not the operation record value (raw data) at the time of interest, but the position where the operation record value at the time of interest is in the distribution of past operation records. It indicates whether to do. The pressure in the blast furnace 1 and the temperature in the blast furnace 1 are used as past operation results.

  As shown in FIG. 6, “in-blast pressure information” of the present technology example is “in-furnace pressure information” indicating how much the pressure situation in the blast furnace 1 is different from the past performance, and in-furnace temperature for each stave unit. "Stave temperature deviation degree information" showing how much the center of gravity of the distribution in the circumferential direction is different from the past actual result, and "Stave temperature deviation degree information for each stave unit" Average temperature information ”.

The in-furnace pressure information is a numerical display of how the current pressure in the blast furnace 1 is in relation to the temporal change in the past pressure in the blast furnace 1 (for example, the past 3 months). Is.
To calculate the in-furnace pressure information, for example, a frequency distribution showing the distribution of past actual pressures in the blast furnace 1 is created, and at which position on the frequency distribution the value indicating the current blast furnace 1 pressure status is set. Whether or not it is located is calculated, and how far the calculated position is from a preset setting value (for example, the median value or average value of the frequency distribution) is calculated.

If the calculated in-furnace pressure information is far from the set value, it indicates that the pressure change in the blast furnace 1 is large, and therefore the in-furnace situation in the blast furnace 1 may become unstable.
In the "in-furnace pressure information (furnace pressure score)" shown on the upper left side of FIG. 6, the in-furnace pressure information is digitized as 0 to 4, and when the in-furnace pressure information is 0 to 2, the blast furnace is It shows that the fluctuation of the pressure in 1 is small, and it is said to be in a normal state.

On the other hand, if the in-furnace pressure information is 2 to 3, it indicates that the fluctuation of the pressure in the blast furnace 1 is rather large, and it is considered that there is a possibility that the blast furnace 1 is deviating from the normal state. When the information is 3 to 4, it indicates that the fluctuation of the pressure in the furnace is extremely large, and it is considered that the inside of the blast furnace 1 may be in an unstable state.
For example, as shown in the upper left part of FIG. 6, the in-furnace pressure information at the staves of B3, A1, and A2 each show a score of 4, which indicates that the pressure fluctuation in the blast furnace 1 is extremely large. On the other hand, the in-furnace pressure information at the A5 stave shows a score of 1, indicating that the pressure fluctuation in the blast furnace 1 is small.

In other words, the current "furnace pressure score" indicates that the height from the lower part of the blast furnace 1 to the vicinity of the stave of A2 is in a state deviating from the normal state, so if operation is continued as it is, there is a possibility that blow-through may occur. There is.
For example, even if the value of the blow-through occurrence prediction score or the blow-through scale prediction score is low, if the current “furnace pressure score” is large, as shown in the upper left part of FIG. It turns out that you need to be careful.

In this way, the blast furnace information plays an important role of correcting and complementing information such as a blow-through occurrence prediction score and a blow-through scale prediction score.
Now, in the middle part on the left side of FIG. 6, the deviation degree information of the stave temperature, which is one of the information in the blast furnace, is shown. The deviation degree information of the stave temperature indicates how much the in-furnace temperature for each stave unit differs from the past performance.

In calculating the deviation degree information of the stave temperature, for example, the stave of A2, the stave of A3, the stave of A4, the furnace temperature in the circumferential direction of each blast furnace 1 are measured, and the radar distribution chart of the measured furnace temperature is shown. think of. The center of gravity position of the reactor temperature data in this radar distribution map is obtained, and the deviation (temperature deviation) between the obtained center of gravity position and the center position of the radar distribution map is obtained.
After that, create a frequency distribution of past results of temperature deviation, figure out at which position on the frequency distribution the current temperature deviation value is located, and the calculated position is the preset setting. How far from the value (for example, the median or average of the frequency distributions) is calculated.

If the calculated temperature deviation is far from the set value, the temperature deviation in the circumferential direction in the blast furnace 1 is large, and an abnormality such as a high temperature in one circumferential direction may occur. There is a nature.
Generally, the high temperature gas rises in the central part of the furnace, but when the internal conditions of the furnace become unstable, a “uniform flow” occurs in the peripheral part, and the wall temperature in the specific ambient direction may increase due to the influence. When the deviation of the stave temperature is large, there is a possibility that such gas flow turbulence and destabilization occur.

In the "Stave temperature deviation degree information (stave temperature deviation degree score)" shown in the middle part on the left side of FIG. 6, the stave temperature deviation degree information is digitized as 0 to 4, and the stave temperature deviation degree information is 0. In the case of up to 2, the deviation of the stave temperature is small, and it is said that the normal operation is performed.
On the other hand, if the stave temperature deviation degree information is 2 to 3, it indicates that the stave temperature deviation is rather large, and it is considered that an abnormality may be occurring in the blast furnace 1, and the stave temperature deviation degree is high. When the information is 3 to 4, the deviation of the stave temperature is extremely large, for example, it indicates that the temperature is high only in one circumferential direction of the blast furnace 1, and the gas flow is disturbed in the blast furnace 1. It is said that there is a possibility.

For example, as shown in the middle part on the left side of FIG. 6, the degree of temperature deviation in the stave A2 is about 1.8, and the deviation of the stave temperature at the present stage is small, but there is a tendency to deviate from the normal state. It is shown that.
Further, since the degree of temperature deviation in the A3 stave and the temperature in the A4 stave are both less than about 1, the deviation of the stave temperature is small, indicating that the inside of the furnace is within the range of normal operation.

  Under such a circumstance, even if the values of the blowout occurrence prediction score and the blowout scale prediction score are low, as shown in the upper left part of FIG. 6, the temperature deviation degree in the stave of A2 is slightly large. , It turns out that the situation requires attention. That is, the blast furnace information has an important role of correcting and complementing information such as a blow-through occurrence prediction score and a blow-through scale prediction score.

The average temperature information of the stave, which is one of the information in the blast furnace, shows how much the average temperature in the furnace for each stave unit is different from the past actual result.
In calculating the average temperature information of the stave, first, the average temperature in the furnace for each stave unit is calculated. In addition, a frequency distribution of past actual results of the average temperature inside the furnace for each stave is created, and at which position on the frequency distribution the current average temperature inside the furnace of the stave is located is calculated. , How far from a preset setting value (for example, the median value or average value of the frequency distribution) is calculated.

If the calculated in-furnace pressure information is farther than the set value, it indicates that the change in the average in-reactor temperature of the stave is large, so there is a possibility that an abnormality has occurred in the blast furnace 1.
In the "Stave average temperature information (Stave average temperature score)" shown in the lower left part of FIG. 6, the current average temperature of the stave is digitized as 0 to 4, and the average temperature information of the stave is 0 to 2. In the case of, the difference between the average temperature of the stave and the set value is small, and it is said that normal operation is performed.

  On the other hand, when the average temperature information of the stave is 2 to 3, it indicates that the difference between the average temperature of the stave and the set value is slightly large, and it is possible that deviation from the normal state is occurring in the blast furnace 1. If the average temperature information of the stave is 3 to 4, it indicates that the difference between the average temperature of the stave and the set value is extremely large, indicating that the blast furnace 1 deviates from the normal state. ing.

For example, as shown in the lower left part of FIG. 6, the information on the average temperature in the furnace in the stave of A2 shows about 1.4, and the difference between the average temperature of the stave and the set value is small, and It is within the range of operation.
On the other hand, the information on the average temperature of the stave of A3 and the stave of A4 both show about 1.8, and the difference between the average temperature of the stave and the set value is rather large. And the difference between the set value and the set value becomes large.

In other words, the current “average temperature information of stave” indicates that the blast furnace 1 is in the range of normal operation but needs attention, so for example, the blow-through occurrence prediction score and the blow-through scale prediction score. Even if the value shows a small value, it is advisable to prepare to change various operating conditions and the like in preparation for a blow-by.
From the above, information indicating the possibility of blow-through can be obtained from the “in-furnace information” including the furnace pressure information displayed on the prediction monitor 11, the stave temperature deviation degree information, and the average temperature of the stave. it can. Based on this information, various operating conditions can be controlled so that the blast furnace 1 can operate stably.

The time series data and blast furnace information displayed on the prediction monitor 11 of the present technology example are stored in the computer 12 or the like provided in the prediction system 10, and are indicated by arrows in the lower side of FIG. By operating the icon and setting it to a desired time, each time series data and the blast furnace information at that time may be extracted.
As described above, by displaying the time series data of the “blown-out scale prediction score”, the time series data of the “blown-out occurrence prediction score”, and the “inside blast furnace information” on the prediction monitor 11, the blast furnace 1 At the time of operation, the possibility of occurrence of blow-through and the prediction of the blow-through scale will be shown as a highly reliable index.

Therefore, the operator M can detect the sign of a blast furnace trouble such as blow-through early by visually observing the prediction monitor 11 of the present technology, can appropriately change the operating condition, and reliably prevent the blow-through. It becomes possible.
"Second technology example"
Next, the operating condition evaluation system of the second technical example will be described.

The operation status evaluation system of the second technical example displays whether or not the blast furnace is in a non-steady state (unsteady state) as in the first technical example. It is the one that displays the scale (hereinafter sometimes referred to as the unsteady monitor).
That is, when the furnace condition of the blast furnace is poor, there occurs a phenomenon in which the hot air blown from the tuyere passes through the end of the blast furnace and becomes uneven flow. If such a nonuniform flow occurs, heat exchange will not be much done inside the blast furnace, and hot air will flow to the middle part, upper part, and furnace top of the blast furnace, and the middle stave temperature, upper stave temperature, and furnace top gas temperature will rise. . In particular, in a blast furnace in which coke is mainly charged, the blown hot air has a small proportion of coke and passes through the end portion of the blast furnace, which has poor air permeability, resulting in a large pressure loss. The inventors of the present invention have clarified through analysis that a series of flows leading to the above-mentioned uneven flow occurs before the occurrence of a serious blow-through.

  However, with the conventional method, it was difficult to evaluate a series of flows leading to such a drift. For example, in the conventional blast furnace operating method disclosed in Japanese Unexamined Patent Publication No. 2006-111904, a copper stave cooler is provided in the blast furnace, and the frequency at which the furnace body temperature measured by the stave cooler exceeds a certain value is calculated, and the frequency is calculated. If the value exceeds the control value, it is judged that the risk of blow-by is large.

  When the above-mentioned non-uniform flow occurs, the temperature in the surrounding specific direction becomes significantly higher than in the other directions, so that not only the temperature but also the balance changes. For example, when the temperature of the furnace body as a whole is rising, but when the wind is flowing stably in the center, it cannot be said that there is uneven flow, and the risk of blow-through is not high. . However, in the conventional blast furnace operating method described above, there is a possibility that it may be judged that the risk of blow-through is great even in such a steady state. That is, this conventional blast furnace operating method cannot reliably capture a series of blow-through flows, and it is difficult to accurately predict blow-through.

  Further, in Japanese Unexamined Patent Publication No. 5-2717228, a plurality of pressure sensors are evenly arranged in the circumferential direction of the shaft portion of the blast furnace, and these pressure sensors are arranged in a plurality of steps in the height direction. A method for preventing blow-through by measuring the pressure with a plurality of these sensors is disclosed. In this method, among the measured values measured by multiple sensors arranged at the same height, the difference between the maximum value and the minimum value is obtained, and the obtained difference is the average pressure value of the sensors of the same height. When the ratio exceeds a predetermined ratio, the wind is reduced because there is a risk of blow-by.

  However, according to the findings of the present inventors, the technique of the above-mentioned Japanese Patent Laid-Open No. 5-2717228 predicts blow-through by using only fragment information of non-uniform flow represented by furnace body pressure. In many cases, the furnace body pressure balance fluctuates even in a steady state, and if the flow is not biased to the extent that there is a high risk of blow-through, the technology described above makes a false decision that the risk of blow-through is high. There is a possibility that it will end up.

Therefore, in view of the above-mentioned problems of the conventional technology, in the operation status evaluation system of the second technology example, not only the furnace body pressure and the stave temperature but also a plurality of indexes such as the blast pressure and the furnace top gas temperature are comprehensively evaluated. In combination, the risk of blow-through is determined to prevent leakage.
That is, as shown in FIG. 7 to FIG. 17, the operation status evaluation system of the second technical example uses the step of acquiring data from the blast furnace and the furnace body pressure and the stave temperature in the data to determine the poor ventilation condition. The step of obtaining the index value that represents, the step of obtaining the score that indicates whether or not a series of flows that occur before the blowout has occurred using the blast pressure and the furnace top gas temperature in addition to the furnace body pressure and the stave temperature, and the index value and the score Is used to determine the risk of blow-through when the two values are different from the normal value.

  The operation state evaluation system of the second technical example adopts such a configuration because the furnace body pressure, the stave temperature, the blast pressure, and the furnace top gas temperature change along a series of flows before the blow-by. This is because it is known empirically. In other words, the operation status evaluation system of the second technical example captures all of this series of changes, so that there is no erroneous determination as compared with the conventional method, and therefore a more accurate blow-through prediction is achieved, and production volume is not reduced and blow-through is also possible. It enables stable operation that is unlikely to occur.

Next, the index value and the score used in the judgment of the blow-through in the operation status evaluation system of the second technical example, and the judgment of the blow-through using these will be described.
The index value described above is a numerical value representing the poor ventilation state of the blast furnace, and is a combination of the furnace body pressure index value, the center of gravity index value, and the temperature difference index value for evaluating different states of the blast furnace. That is, in the operation status evaluation system of the second technical example, in the first step, the fluctuation of the furnace body pressure at a plurality of heights over a long period of time and the fluctuation over a short period of time are used as the furnace body pressure index value. Ask. This furnace body pressure index value is a numerical value indicating the poor ventilation state of the blast furnace, which appears in the furnace body pressure. When the furnace body pressure index value is obtained in the first step as described above, how much the stave temperature deviates in the circumferential direction is obtained in the second step as a barycenter index value. This center-of-gravity index value is a numerical value indicating the poor ventilation state of the blast furnace that appears in the temperature distribution of the stave temperature along the horizontal direction. Further, when the barycentric index value is obtained in the second step, the difference in temperature between the upper part and the middle part of the blast furnace is obtained as a temperature difference index value in the third step. This temperature difference index value is a numerical value indicating the poor ventilation state of the blast furnace, which appears in the temperature difference of the stave temperature along the vertical direction of the blast furnace. In this way, when the three index values (the furnace body pressure index value, the center of gravity index value, and the temperature difference index value) are obtained in the first to third steps, the minimum value of these three index values (three indexes The smallest value) is adopted as the index value (blowing prediction index value) that represents the poor ventilation condition. Finally, it is determined whether the adopted blow-out prediction index value is equal to or larger than a preset set value, and the result of this determination is used to determine whether there is a risk of blow-through.

  If the operating condition evaluation system of the second technical example described above is used, most blow-throughs can be detected, equipment damage due to blow-throughs can be prevented, and molten steel productivity can be secured. That is, in general, when a blow-by occurs, immediately before that, the state in which the furnace body pressure fluctuates both in a long time span and in a short time span, and the distribution of the stave temperature in the circumferential direction is biased. In many cases, the temperature in the upper part of the blast furnace rises and there is no temperature difference. On the contrary, if there is only one of these states, there are many cases where even the unstable situation does not lead to the void. In order not to overlook and to predict the blow-through without excess, obtain three index values that represent each of the three states, and even the smallest of these index values will be greater than or equal to the set value, in other words, all indexes. When the value exceeds the set value, it is judged that the ventilation state deteriorates and there is a risk of blow-by.

  The above-mentioned score indicates whether or not a series of abnormal states, particularly a series of abnormal states known to occur before the blow-through, actually occur. This score is a combination of, in other words, a combination of five scores, that is, a furnace upper temperature score, a furnace middle temperature score, a specific direction protrusion score, a blast pressure score, and a furnace top gas temperature score. There are various processes leading up to the atrium, and judgments that rely on only one particular score cannot reduce oversight or overjudgment. Therefore, the use of a plurality of scores as described above is used to predict the occurrence of blow-through.

  Of these five intermediate scores, the furnace upper temperature score is a score that becomes an abnormal value when the temperature of the upper part of the blast furnace exceeds a reference value. Further, the middle-furnace high temperature score is a score that becomes an abnormal value when the stave in the middle part of the blast furnace exceeds a reference value. Before the blow-through, the upper part of the blast furnace and the middle part of the blast furnace are often at a higher temperature than in normal times, so the upper part temperature of the upper part of the furnace and the higher part temperature of the middle part of the furnace are obtained as a score of these abnormal states.

  The specific direction protrusion score described above is a score that becomes an abnormal value when the blast furnace is biased. The blast pressure score is a score that becomes an abnormal value when the pressure loss exceeds the reference value. In other words, before the blow-through, the flow of hot air inside the blast furnace becomes uneven and the pressure loss becomes larger than usual, so the specific direction protrusion score and the blast pressure score are obtained as a score of these abnormal conditions. Has become.

Finally, the furnace top gas temperature is judged to be an abnormal value when the furnace top gas temperature exceeds the reference value. In other words, the furnace top gas temperature score is higher than the normal time before the blow-through, so the furnace top gas temperature score should be calculated as an abnormal condition that the temperature of the furnace top gas is high. Has become.
Regarding the five scores thus obtained, an abnormal state (precursor state) that occurs before the blow-through significantly occurs only when a predetermined number of the five scores have an abnormal value. It is judged that there is a danger of blow through. The number of scores judged to be at risk of blowout may be one, or may be two or more. Alternatively, the judgment may be made by preferentially picking up only a specific score among the five scores.

For example, in the case of the flowchart in FIG. 9, when the furnace upper temperature score shows an abnormal value, and if two or more of the remaining four scores have an abnormal value, the It is judged that there is a risk of "and displayed on the display.
By evaluating a series of abnormal states known to occur before such blow-through as a score, it becomes possible to accurately predict the risk of blow-through. For example, when the wind is increased in a steady operation, it may be erroneously determined that blow-by will occur only by the stave temperature and the furnace body pressure. However, as in the second technical example, if a blow-through is judged based on whether or not a plurality of scores are bad at the same time, it is rare that a plurality of scores become abnormal values at the same time even when the wind is blowing up. It is possible to reduce the number of over-detections and prevent a decrease in the production amount due to the action for preventing blow-through (action of reducing wind).

As shown in FIG. 9, when the above-described index value and score are obtained, the risk of blow-through is determined using the index value and the score when the two values are different from the normal value. This determination is performed by the above-described prediction unit, and when there is a risk of blow-by, the display unit displays it.
Next, the procedure of determining the risk of blow-through performed by the prediction unit will be described with reference to FIGS. 9 to 17.

  Note that FIG. 9 shows the overall processing flow of the blow-through prediction in the second technical example, and FIGS. 10 to 12 show the processing flow in detail when each of the three index values is calculated. 13 to 16 show in detail the processing flow when each of the five scores is calculated. Further, FIG. 11 is a graph for explaining the standardization process used when comparing the magnitudes of the three index values.

  As shown in FIG. 9, first, operation data such as the furnace body pressure and the stave temperature of the blast furnace are acquired by a plurality of sensors provided in the circumferential direction and the height direction of the blast furnace (S9-1). The acquired operation data is transmitted to the analysis unit (prediction unit) (S9-2). The predicting unit performs data shaping such as removing an abnormal value of the operation data acquired by using filtering or the like, and data shaping such that a period in which the operation data is not acquired is substituted with the immediately preceding operation data value ( S9-3). Since this data shaping process is for facilitating the process described later, it does not have to be performed when the operation data is not missing or when there is no abnormal value. Three index values (intermediate index values, that is, furnace pressure index value, center of gravity index value, and temperature difference index value) are calculated from the operation data after performing such shaping processing (S9-4), and calculation A blow-out prediction index value is calculated from the calculated index value (S9-5). In addition, from the operation data after performing the shaping process, five scores that tend to be large before the blow-through (upper furnace upper temperature score, middle furnace upper temperature score, specific direction protrusion score, blast pressure score, and furnace top The gas temperature score) is calculated (S9-6).

The details of the process (S9-4) for calculating the above-mentioned three index values will be described with reference to FIGS. 10 to 12. That is, FIG. 10 shows a flow for calculating the furnace body pressure index value out of the three index values, FIG. 11 shows a flow for calculating the center of gravity index value, and FIG. 12 shows a flow for calculating the temperature difference index value. Is.
As shown in FIG. 10, when calculating the furnace body pressure index value, first, the operation data of the furnace body pressure is acquired by a plurality of sensors provided in the height direction and the peripheral direction (circumferential direction) of the blast furnace (S10- 1). As for the operation data of the furnace body pressure, the average value of the operation data in the circumferential direction is calculated between the sensors located at the same height. The average value of such circumferential operation data is obtained for each height of the blast furnace (S10-2). Further, the average value of the above-mentioned operation data is calculated as a variance value for the past fixed period (for example, a variance value for 120 minutes) (S10-3). Further, a plurality of time average values of the dispersion values (for example, four average values of the past 5 minutes, 60 minutes, 120 minutes, and 240 minutes) are calculated (S10-4). A binary score (for example, 0 point or 1 point) is calculated depending on whether or not the time average value calculated above is a set value or more (S10-5). Using this score at each height, a score representing the degree of fluctuation of the furnace body pressure is calculated for each height (S10-6). For example, assuming that the binary score is 0 or 1, calculation is performed by the following formula (1).

The binary score shown in the above equation (1) is calculated for all height levels, and they are summed (S10-7), and the sum is standardized (S10-8). Is the furnace body pressure index value.
Note that “normalization” used when calculating the three index values means setting two values (Pmin and Pmax in the figure) for a certain value (in this case, the total value), as shown in FIG. However, it is to correspond to a value of 0 to 1 that linearly rises only in the meantime (hereinafter, the same in the present specification). This normalization is performed for the purpose of relatively comparing the magnitudes of the three index values, and if the purpose can be achieved, there is no problem even if another standardization method is used.

  As shown in FIG. 11, when calculating the center-of-gravity index value, first, the stave temperatures at a plurality of heights and the circumferential direction of the upper part of the blast furnace are acquired (S11-1). The measured stave temperature values thus obtained are plotted on a two-dimensional plane as a circular graph. That is, the measured value of the stave temperature is plotted with the radial distance from the center of the blast furnace as the origin (S11-2). In this way, as the stave temperature becomes higher, it is possible to obtain a circular graph in which points indicating measured values are plotted at positions separated from the origin by the temperature. When a plurality of measurement values located at the same height are plotted in this way, the distance between the center of gravity of the points indicating the plurality of measurement values and the center of the blast furnace (the origin of the circular graph) is calculated (S11-3 ). The calculated distance from the center of gravity is calculated by the following equation (2).

  Note that “n” in the equation (2) is the number of temperatures acquired at that height (the number of sensors located at a certain height of the blast furnace), and “i” is the index of the sensor at that height. (i = 1,2 ... n). Further, “Ti” is the temperature acquired by the i-th sensor, and “ei” represents a vector of length 1 having a direction toward the i-th sensor. This distance is standardized (S11-4), and the maximum value of the total height (S11-5) is the center-of-gravity index value.

  As shown in FIG. 12, similarly, the stave temperature at a plurality of heights in the upper part and in the circumferential direction is acquired (S12-1). After taking a time average for this temperature for a certain period in the past (for example, 120 minutes) (S12-2), an average value in the circumferential direction is calculated for each height (S12-3). Of the heights corresponding to the average value obtained here, the difference between the bottom value and the top value (often a negative value) is calculated (S12-4), and this value is normalized. What was done (S12-5) is the temperature difference index value.

  Finally, the minimum value is selected from the index values obtained by the above-mentioned procedure, that is, the furnace body pressure index value, the center of gravity index value, and the temperature difference index value. Since the furnace body pressure index value, the center of gravity index value, and the temperature difference index value are all standardized to have numerical values of 0 to 1, the magnitudes can be compared. The minimum index value thus selected becomes the blow-out prediction index value (S9-5).

Along with the calculation of the blow-out prediction index value (index value indicating poor breathability) described above, a score indicating poor furnace conditions is also calculated (S9-6). Any of these five scores is judged to be an abnormal value if it is equal to or greater than a preset set value.
Next, the calculation procedure for each score will be described.
As shown in FIG. 13, when calculating the furnace upper temperature score, first, a plurality of stave temperatures are acquired in the height direction and the circumferential direction of the upper part of the blast furnace (S13-1). For the stave temperature measured by each sensor, a time average for the past fixed period (for example, 120 minutes) is calculated (S13-2). With respect to the time average of the sensors with the same height of the blast furnace, the average value in the circumferential direction is calculated (S13-3). An average value is obtained for each height of the blast furnace and standardized from 0 to 1 (S13-4). The average value (S13-5) of this normalized value and the temperature difference index value obtained in S4-35 is the furnace upper temperature score.

  As shown in FIG. 14, when calculating the furnace middle temperature increase score and the specific direction protrusion score, first, the stave temperature is acquired at a plurality of heights and in the circumferential direction (S14-1), and this is calculated by the following equation (3). ) For normalization (S14-2).

Among the normalized values calculated from the stave temperature measured by the sensor located at the height of the middle part of the blast furnace, the maximum value is the middle part furnace high temperature score (S14-3). Further, the difference between the maximum value and the minimum value of the normalized values is calculated in each circumferential direction (S14-4), and the maximum value is the specific direction protrusion score (S14-5).
As shown in FIG. 15, when the blast pressure is acquired (S15-1), the blast pressure score is calculated as the maximum value in a certain past period (for example, 120 minutes) (S13-2).

As shown in FIG. 16, the furnace top gas temperature score is calculated as the maximum value in the past certain period (for example, 60 minutes) in all directions when the furnace top gas temperatures are acquired in a plurality of ambient directions (S16-1). It is done (S16-2).
The blow-through prediction index value obtained by the above-mentioned procedure and the five scores are used to determine the risk of blow-through (S9-7 to S9-11).

Specifically, as shown in FIG. 9, it is determined whether the blowout prediction index value obtained in (S9-5) is equal to or greater than the set value (S9-7). If it is equal to or greater than the set value, proceed to (S9-8) and judge the risk of blow-through based on the poor reactor conditions. If it is less than the set value, proceed to (S9-11), judge that there is no danger of blow-through, and do not display on the display.
Next, it is determined whether the furnace upper temperature score is an abnormal value (S9-8). If the obtained blow-out prediction index value is greater than or equal to the set value, proceed to (S9-9). However, if it is less than the set value, the process proceeds to (S9-11), it is determined that there is no danger of blow-through, and the display on the display unit is not performed.

  Finally, out of the remaining four scores (reactor middle part temperature rise score, specific direction protrusion score, blast pressure score, and furnace top gas temperature score) obtained in (S9-6), three or more are abnormal values. It is determined whether there is any (S9-9). If there are three or more, proceed to (S9-10), determine that there is a risk of blow-through, and display the information on the display unit. However, if the number is two or less, the process proceeds to (S9-11), it is determined that there is no danger of blow-through, and the display is not displayed.

Note that in S11-3, the position of the center of gravity is calculated in order to check the deviation of the stave temperature at each height, and for example, the deviation of the temperature is indicated by using the difference between the maximum value and the minimum value of the circumferential stave temperature. It may be replaced with the calculation method of.
In addition, the normalization performed in S14-2 is a process for comparing respective values of the staves being measured (materials, etc.), and the staves to be compared have the same conditions. Or, if there is no problem in comparison, normalization need not be performed.

Furthermore, the above-mentioned blast pressure score and furnace top gas temperature score are scores indicating that the respective data values are high (being large), so if an increasing tendency can be shown, the maximum value for the past certain period will be It may not be necessary, and may be an average value or a median value in a certain past period.
The display content displayed on the display unit may show not only the binary information indicating whether there is a risk of blow-by but also the value of the calculation process. For example, the calculated blow-through prediction index value and five scores may be directly displayed on the display unit without performing the determination procedure of the processes C9-5 to C9-7.

Further, the judgment result and the calculation result do not have to be displayed on the display unit. In that case, the data of judgment results and calculation results will be used as data for automatically changing operating conditions and for later analysis and discussion. For example, if data on whether or not there is a risk of blow-through is acquired over a long period of time, it is possible to easily extract the period during which the state of the blast furnace was unstable, dating back to the past. Can be used to analyze the relationship between.
"Third technology example"
Next, the operation status evaluation system of the third technical example will be described.

  As shown in FIG. 18, the operation status evaluation system of the third technology example, like the second technology example, analyzes the operation data and evaluates the risk of blow-through. Not only is it determined, but the operation operator is instructed when to take action to avoid blow-through. In other words, the operation status evaluation system of the third technical example is a system that determines the timing of performing an action for avoiding a blow through.

For example, Japanese Patent Laid-Open No. 2006-28576 discloses numerical values for predicting fluctuations in the in-furnace pressure based on the pressure and the rate of change over time by measuring the in-furnace pressure with a plurality of pressure sensors provided in the blast furnace shaft. Disclosed is a method and apparatus for predicting in-reactor pressure fluctuations, which displays an alarm when the calculated value exceeds a threshold.
With this technique related to prediction of pressure fluctuations in the furnace, it is possible to display an alarm to the operating operator to notify that abnormal fluctuations in the furnace pressure have occurred. However, since operations in blast furnaces are often empirical and there are individual differences among operating operators, the operating operators who receive the alarm do not always take appropriate actions, and belong to actions to prevent blow-through. Human error is possible. In other words, even if it is possible to accurately predict the occurrence of pressure fluctuations before the blow-through, it is up to the operator or staff to decide when to take action just by displaying the alarm, and to prevent the blow-through. There is a risk that adequate actions will not be taken sufficiently.

Further, like the above-described Japanese Patent Laid-Open Publication No. 5-2717228, when a blow-through is predicted using only fragment information, which is represented by the pressure in the furnace body, that is, the blow-through prevention method disclosed in Japanese Unexamined Patent Application Publication No. 5-27172. There is a possibility that the risk of blow-through may be erroneously determined to be large even in a steady blast furnace state where the condition is not bad, and there is a great problem in the accuracy of predicting the risk of blow-through.
Therefore, in view of the above-mentioned problems of the conventional technology, the operation status evaluation system of the third technology example displays the timing at which an action such as a wind reduction is performed to determine whether or not the final action should be performed. It is possible to present to.

Further, the operation status evaluation system of the third technical example predicts the risk of blow-through by comprehensively using the furnace body pressure, stave temperature, blast pressure, furnace top gas temperature, etc. to predict the risk of blow-through. The accuracy can be greatly improved.
In other words, the operating condition evaluation system of the third technical example uses the “risk of blow-through” determined by the procedure described in the second technical example and the “instantaneous value of furnace pressure” newly introduced in the third technical example. By combining with, it becomes possible to predict the risk of blow-through without overdetection, and to appropriately determine the timing of actions such as wind reduction.

  Specifically, the operation status evaluation system of the third technical example includes a step of acquiring data from the blast furnace, and a step of obtaining an index value indicating poor ventilation condition by using the furnace body pressure and the stave temperature in the data. In addition to the furnace body pressure and stave temperature, the blast pressure and furnace top gas temperature are used to obtain a score that indicates whether or not a series of flows that occur before the blow-through has occurred, and two values are normally used using the index value and score. And a step of calculating the risk of blow-through so that the value becomes higher when the value is different from. Note that these steps are the same as those in the second technical example described above.

The third technical example is different from the second technical example in that it is calculated from the risk and the furnace body pressure in the step of obtaining the furnace body pressure calculation value that increases from the instantaneous furnace body pressure value in the data before the blow-through. When both of the above values exceed the respective set values, the step of determining the wind reduction timing is performed.
According to the operation status evaluation system of the third technical example as described above, it is possible to perform an operation that does not have a personality, does not reduce the production amount as much as possible, and does not cause blow-through. In other words, if only the indication that the reactor is in poor condition is indicated by a display, etc., the operator's judgment remains personal, so the timing of actions after receiving the indication may vary depending on the person, and in some cases it may be appropriate. There is a possibility that a mistake (misoperation) will occur because such an operation is not performed.

However, with the operation status evaluation system of the third technical example described above, more accurate blow-through prediction is performed, and the timing of actions such as wind reduction is appropriately determined based on the prediction result. It is possible to reliably prevent the occurrence of blow-through without lowering the
Hereinafter, the "step of obtaining the calculated value of the furnace body pressure" and the "step of determining the wind reduction timing", which are the features of the operation status evaluation system of the third technical example, will be described.

  The calculated value of the furnace body pressure is the measured value of the furnace body pressure measured by multiple sensors installed in the middle part of the blast furnace (intermediate height between the upper part and the lower part), and the furnace measured by the sensor installed on the top of the furnace. It is the largest of the pressure difference (pressure loss) from the top pressure. That is, the blast furnace provided with the operation status evaluation system of the third technical example has a structure in which "upper", "middle", and "lower" are provided under the furnace top. A plurality of sensors are provided side by side in the circumferential direction at the same height at the upper, middle and lower parts of this blast furnace. Among the plurality of sensors, the measured value of the furnace body pressure measured by each of the sensors existing in the middle portion in the height direction is the furnace body pressure instantaneous value.

On the other hand, the top of the blast furnace is also provided with a plurality of sensors in the circumferential direction to measure the top pressure. The pressure difference between the furnace top pressure measured by the furnace top sensor and the aforementioned furnace body pressure instantaneous value is the furnace body pressure calculated value. The calculated furnace body pressure value can be obtained for all instantaneous furnace body pressure values measured by a plurality of sensors provided in the circumferential direction.
In the case of a blast furnace in which the furnace top pressure is controlled to be constant, the furnace body pressure instantaneous value itself may be used as the furnace body pressure calculated value. Further, when the positions at which the furnace top pressure and the furnace body pressure instantaneous value are measured are different in the circumferential direction, the maximum furnace top pressure may be used as a representative value in the calculation to calculate the furnace body pressure calculated value.

This calculated value of the furnace body pressure can evaluate the pressure loss of the blast furnace, which is closely related to the occurrence of the blow-through, and by combining it with the evaluation result of the blow-through risk explained in the second technical example, It is possible to suppress overdetection.
When the above-mentioned furnace body pressure calculated value is calculated, the calculated furnace body pressure calculated value and the risk of blow-through obtained in the above-mentioned second technical example are used to determine the risk of blow-through or the risk degree described later. Display and the timing of wind reduction.

That is, when the obtained risk of blow-through is equal to or higher than the set value, the risk is displayed to the operating operator, and when the above-mentioned furnace body pressure calculated value is equal to or higher than the set value, the timing for reducing the wind is displayed. To be displayed on the department.
Next, a procedure for determining action timing such as wind reduction performed by the prediction unit will be described with reference to FIGS. 20 and 21.

Note that FIG. 20 shows a procedure for determining action timing such as wind reduction in the third technical example, and FIG. 21 shows a processing flow in detail for calculating a furnace body pressure calculated value. .
As shown in FIG. 20, first, temperature and pressure data are acquired from the top of the blast furnace to the bottom of the blast furnace as operation data of the blast furnace by a plurality of sensors provided in the height direction and the circumferential direction of the blast furnace (S20-1). The operation data thus obtained is transmitted to the prediction unit (analysis unit) (S20-2). The analysis unit performs data shaping by removing abnormal values (such as noise) from the acquired operation data, and substituting / interpolating the data value missing period with the immediately preceding data value (S20-3). This data shaping is to facilitate the subsequent processing when there is a part missing in the operation data. If there is no missing, or if there is an abnormal value or noise, it does not cause a problem. In this case, it is not necessary to format this data. After such shaping of the data, the risk of blow through (S20-4) and the calculated value of the furnace body pressure (S20-5) are calculated from the shaped data.

Since the risk of blow-through is calculated according to the second technical example described above, detailed description thereof will be omitted here.
As shown in FIG. 21, when calculating the furnace body pressure calculation value, an arbitrary height is selected from the height direction of the blast furnace. Then, the furnace pressure instantaneous value at the selected height is acquired by a plurality of sensors provided in the circumferential direction (S21-1). Regarding the height of the furnace body pressure to be selected at this time, the prediction timing is delayed if it is in the upper part, and if it is in the lower part, the influence of disturbance is large and the accuracy deteriorates, so it is better to select it from the middle part of the blast furnace.

Next, the difference (pressure loss) between the furnace top pressure and the furnace body pressure is calculated at each position (each direction) in the circumferential direction (S21-2). The difference between the furnace top pressure and the furnace body pressure thus obtained is the furnace body pressure instantaneous value. When the furnace body pressure instantaneous value at each position in the circumferential direction is calculated in this way, the maximum value is selected from the calculated furnace body pressure instantaneous values, and this is made the furnace body pressure calculated value (S21- 3).
If the blast furnace is controlled so that the furnace top pressure is constant, the furnace body pressure instantaneous value itself may be used instead of the pressure loss. When the positions of the furnace top pressure and the furnace body pressure are different in the circumferential direction, the maximum furnace top pressure (maximum value of the furnace top pressure) may be used as a representative value for calculating the furnace body pressure calculation value.

As shown in FIG. 20, it is confirmed whether or not the obtained risk level is equal to or higher than the set value, and if it is equal to or higher than the set value, the operation unit is displayed (presented) to the operating operator through the display unit. If the “furnace condition is dangerous” is presented first, it is possible to call the attention of the operating operator.
Next, it is judged whether the calculated furnace body pressure value is above the set value, and if the calculated furnace body pressure calculated value is above the set value, the timing of actions such as wind reduction is displayed to the operating operator, or Automatically reduce wind. The action timing and the amount of wind reduction may be a predetermined value or may be changed according to the degree to which the calculated furnace body pressure calculated value deviates from the set value.

In addition, the set values that determine the degree of danger and the calculated value of the furnace body pressure for abnormalities (unsteadiness) are appropriate for the final action timing that can prevent troubles in light of past blow-through cases. It is recommended to use a value adjusted to the timing.
Instead of the degree of risk described above, the “blowing-out occurrence prediction score” obtained in the first technical example may be used. When this "blown-out occurrence prediction score" is used, the risk of blow-through is small, but the value of the "blown-out occurrence prediction score" is high even when the reactor condition is poor. Therefore, the occurrence of blow-through can be prevented more reliably, and the operation of the blast furnace can be performed more safely.

When the risk level fluctuates well, the maximum value in the past fixed period may be used as the risk level for the determination instead of the instantaneous value.
Actions other than wind reduction may be used as actions for preventing blow-through. For example, if the coke ratio is increased, it is possible to reliably improve the furnace condition without fear of disturbing the inside of the furnace due to wind reduction, though the immediateness and quick effect are lacking.
"Fourth technology example"
Next, the operating condition evaluation system of the fourth technical example will be described.

The above-described first to third technical examples are provided with a step of obtaining an index value indicating a ventilation state and a step of evaluating the occurrence of a precursory state of blow-through by a score. On the other hand, the operating condition evaluation system of the fourth technical example relates to the step of obtaining the index value indicating the ventilation state.
That is, it is considered that the above-mentioned “blowing” is caused by the disturbance of the gas flow inside the blast furnace and the deterioration of the air permeability. Operational actions such as detecting such deterioration of air permeability as a sign of blow-by in advance and reducing wind and increasing the coke ratio (operations performed to reduce adverse effects such as heat loss when blow-through occurs). It is the configuration of the "non-steady state monitor" of the first to third technical examples that presents information that serves as a guide for determining.

In the following, in the "unsteady monitor" of the first to third technical examples, the operating condition evaluation system that numerically expresses the risk of blow-through occurrence by utilizing the operating data currently collected will be described. In the following specification, the means for numerically expressing the degree of danger provided in the operation status evaluation system of the fourth technical example may be referred to as a "blown-out alarm".
The “blowing alarm” provided in the operation status evaluation system of the fourth technical example is a phenomenon (state) that is correlated with deterioration of the air permeability of the blast furnace, that is, “the degree of fluctuation of the furnace body pressure increases”, “blast furnace Risk of blow-through occurrence, focusing on phenomena such as "the difference in the stave temperature in the height direction is small in the upper part of the furnace (upper furnace wall)" and "the deviation of the stave temperature measured in the peripheral direction of the blast furnace is large" Is expressed as a numerical value.

  That is, among the above-mentioned phenomena, when the fluctuation of the furnace body pressure is considered, the fluctuation of the furnace body pressure itself occurs due to various factors. Also, the magnitude (level) of fluctuations in the furnace body pressure is an important control item in the operation of the blast furnace, and when the fluctuations in the furnace body pressure actually become large, they are recorded in operation records such as diaries. . When a blow-through occurs, a large fluctuation in the furnace pressure is actually confirmed.

  However, the fluctuation of the furnace body pressure itself frequently occurs other than when the blow-through occurs, and the fluctuation of the furnace body pressure does not always lead to the blow-through. In other words, even if it is meaningful for operation management to quantify the degree of fluctuation of the furnace body pressure itself, if only the degree of fluctuation of the furnace body pressure is paid attention to issue the “blown-through alarm”, it is issued. However, it will unnecessarily increase the excessive number of official announcements that no void will occur.

In other words, fluctuations in the furnace body pressure represent turbulence in the ventilation state of the blast furnace, but turbulence in the furnace body pressure leads to fluctuations in the gas flow inside the blast furnace, and eventually blow-through occurs. In order to understand the process, it is necessary to pay attention not only to the fluctuation of the furnace body pressure but also to the furnace wall temperature above the furnace body, for example, the values of the temperature sensors provided at “A4” to “A6” in FIG.
The furnace wall temperature (temperature data of "A4" to "A6") in the upper part of the furnace body is in a certain temperature range when the blast furnace is in a steady state, and between "A4" to "A6". There is often a constant temperature difference. Further, in “A4” to “A6”, the temperatures (stave temperatures) measured by the temperature sensors provided in the circumferential direction are substantially the same.

  However, when a phenomenon in which the gas flow inside the blast furnace is disturbed and the high-temperature gas suddenly rises along the inner wall of the blast furnace, in other words, a precursory phenomenon of blow-by occurs, as shown in the left side of FIG. The temperature difference in the height direction becomes smaller. ” Further, when the above-described precursory phenomenon of blow-through occurs, as shown on the right side of FIG. 22, there is a phenomenon such that “in the peripheral direction, the temperature of the part where the gas flows becomes higher than that of the other parts”.

Therefore, in the “blowing-out prediction alarm” of the fourth technical example, the risk of occurrence of blowing-through is expressed as a numerical value by taking into consideration whether or not the above-mentioned three phenomena occur.
Specifically, whether or not three phenomena have occurred is expressed as the above-mentioned three index values. This index value is not an absolute value itself such as a temperature difference, but is expressed by normalizing the deviation from the steady state value to a real number of 0 to 1. For example, a frequency distribution of values over a certain period in the past is obtained, and whether or not the value is included in the low frequency side of the obtained frequency distribution, for example, the top 10% in the frequency distribution is used as the threshold value for deviation from the steady state. This makes it possible to set thresholds that match the actual operating conditions of the process.

The specific calculation procedure of the three index values will be described below.
First, the furnace body pressure fluctuation index value will be described. This furnace body pressure fluctuation index value corresponds to the furnace body pressure index value of the second technical example. The furnace body pressure fluctuation index value is obtained from the furnace body pressures measured by four furnace body pressure sensors provided in the blast furnace and having different installation heights. That is, since each installation height has a plurality of sensors in the circumferential direction, the average value of the values measured by these sensors is obtained.

  Next, the variance in each section (time) over the past 30 minutes, 60 minutes, 120 minutes, and 240 minutes is calculated with respect to the obtained average value, and the magnitude of fluctuation at each position is scored in four stages. Express with. This score has four levels from 0 point indicating “small fluctuation” to 3 points indicating “large fluctuation”, and the scores at four positions in the height direction are added together. By doing so, it is possible to represent the fluctuation in the furnace pressure at a level where the total points are 0 to 12. It should be noted that this furnace body pressure fluctuation index value is indicated by the letter "V" in order to distinguish it from other index values.

  The center-of-gravity index value is a numerical representation of the center of gravity of the temperature of the furnace wall (stave temperature), that is, the deviation in the circumferential direction. Specifically, in each of “A4” to “A6” in FIG. 2, the center of gravity position of the measured value in the circumferential direction is calculated based on the measured value of the stave temperature in the circumferential direction, and is calculated from the center position of the blast furnace. The deviation (degree) to the determined position of the center of gravity is calculated. In addition, in order to distinguish this center-of-gravity index value from other index values, the characters “C1”, “C2”, and “C3” are shown.

  The temperature difference index value expresses the temperature difference of the stave temperature in the middle part of the blast furnace. That is, as described above, the average temperature in the circumferential direction is measured for each of “A4” and “A6” in FIG. 2, and the difference in the height direction of the measured average temperature is calculated. The temperature difference between "A4" and "A6" thus obtained is the temperature difference index value. The temperature difference index value is indicated by the letter "T" in order to distinguish it from other index values.

In the above-mentioned precursory state leading to blow-through, the furnace body pressure fluctuation index value increases, the temperature difference index value decreases, and the center of gravity index value increases. Therefore, when all of the furnace body pressure fluctuation index value, the temperature difference index value, and the center of gravity index value exceed a certain level, an alarm is issued as a risk of blow-through.
It should be noted that how to specifically issue the "blown-out prediction alarm" is determined by determining the threshold values of the furnace body pressure fluctuation index value, the temperature difference index value, and the center of gravity index value, which serve as a guide when issuing the alarm. Is determined using the operation record data as follows.

First, in determining the threshold values of the furnace body pressure fluctuation index value, the temperature difference index value, and the barycentric index value, the following three steps are required. Note that, in the following, only the result of the temperature difference index will be described as an example for the sake of simplicity.
The furnace body pressure variation index value, the temperature difference index value, and the center of gravity index value described above cannot be compared because they have different ranges of variation and different scales. Therefore, first, each index value is normalized so that the index values can be compared with each other.

  For example, the average value and the standard deviation of the temperature difference index values are obtained for the operation data acquired in the latest one year period. Then, each data of the temperature difference index value is scaled (normalized) using the obtained average value and standard deviation. The index value subjected to such scaling shows a cumulative frequency distribution as shown in FIG. Note that FIG. 23 shows the maximum stave temperature in the circumferential direction and the center of gravity of the stave temperature in the circumferential direction at each height of S4 to S6, and does not show the temperature difference index value itself. However, since the above-mentioned temperature difference index values also have the same change tendency as these values, the cumulative frequency distribution of these values will be given here instead. The step of obtaining the cumulative frequency distribution of the temperature difference index values after scaling in this way is the first step.

  When the cumulative frequency distribution of temperature difference index values is obtained in the first step described above, the "score function" is defined so that the cumulative frequency rises at 90% and becomes 1 at 99%. Specifically, as shown in FIG. 24, this score function has a score of “0” when the cumulative frequency of the actually measured temperature difference index values is 90%, and a score when the cumulative frequency is 99%. Is a function that becomes "1".

  That is, in the data of the actual temperature difference index value, the cumulative frequency distribution has “unevenness” even if it is monotonically increasing, and it may be difficult to determine the values corresponding to 90% and 99%. Therefore, here, for convenience, the cumulative frequency curve is approximated by, for example, an exponential curve, and then XL and XU are determined (according to the identified equation). The step of determining the score function described above is the second step.

S (V), S (C1), S (C2), and S (C3) are the score functions determined for the furnace body pressure fluctuation index value, the temperature difference index value, and the barycentric index value in the second step. , S (T).
When the score function of each index value is obtained in this way, the integrated score is calculated as in equation (4) using the obtained score function.

The integrated score S obtained by the above equation (4) defines a threshold value common to the furnace body pressure fluctuation index value, the temperature difference index value (stave temperature difference index value), and the center of gravity index value. An alarm is issued when all index values exceed a certain threshold. The integrated score S is given as a real number of 0 to 1.
The center of gravity index value is the maximum value of S4, S5, and S6. This is because the deviation of the temperature distribution in the circumferential direction may be large in any of these.

When the sections in which the integrated score S exceeds 0.8 described above are roughly classified, there are “one-shot sections (one-shot score peaks)” and “swarm sections”. Is often occurring.
It is also considered that when the gas flow in the furnace is disturbed, the temperature in the peripheral direction of the blast furnace becomes uneven, or the temperature fluctuation in the height direction of the blast furnace is increased, and the score is increased. Furthermore, when the wind in the blast furnace is reduced or increased, fluctuations in the furnace pressure also increase, and the score also increases with the abrupt temperature change described above.

As for the state transition pattern (scenario) of the blast furnace leading to blow-through, for example, when the air permeability is poor and the wind is reduced, the hot metal temperature tends to decrease, but the wind was increased to return to a steady state as a measure after wind reduction. In this case, it is considered that the hot metal temperature rises rapidly and the wind pressure increases (the score also rises), leading to blow through.
From the above, it is considered that the risk of blow-through can be determined with high accuracy by using the operation status evaluation system of the fourth technical example. That is, the furnace body pressure fluctuation index value, the temperature difference index value, and the center of gravity index value described above represent how much the furnace body pressure fluctuation and the temperature fluctuation deviate from the steady state, and when these deviations are large, The risk of blowout is also considered to be high. Therefore, when judging the risk of blow-through in actual blast furnace operation, instead of individual index values, a combination of multiple index values should be used to determine the movement of each index value (single shot / swarm) and surrounding conditions (loading pitch). It is considered that it is possible to predict the blow-through with high accuracy by displaying the "deviation from the steady state" together with (or the occurrence of increased / decreased wind).
"Fifth technology example"
Next, the operating condition evaluation system of the fifth technical example will be described.

  The above-mentioned fourth technical example relates to the "blown-out prediction alarm" that detects a "normally extremely rare state (state that cannot occur)" and notifies the operating operator. In other words, the "blowing prediction alarm" of the fourth technical example is "the furnace body pressure time variation in the middle, upper part, and top of the blast furnace", "upper stave (S4 to S6) temperature (1 minute interval) in the circumferential direction. Deviation (center of gravity deviation) "," temperature distribution in the height direction of the blast furnace (temperature difference between S6 and S4) ", and monitor" increase in furnace pressure fluctuation when air permeability deteriorates "or" surrounding the blast furnace " The non-uniformity of stave temperature distribution in the direction is used to judge the risk of blow-through.

  However, there are various causes and circumstances leading to the void, and the time transitions and values of the indicators and values described so far differ depending on the difference in such phenomena. In the method of the fourth technical example, in order to minimize the “excessive announcement” and the “missing”, a plurality of indicators and scores are combined under various conditions to devise a method capable of dealing with the variety of phenomena. However, it is inevitable that a certain percentage of oversights and over-issues will occur. For example, “excessive announcement” means that even if a blow-through alarm is issued (even if the total score S (blowing prediction score) of the fourth technical example exceeds a certain threshold), no blow-through actually occurs. For example, it is likely to occur at the time of start-up of a breeze or when changing the flow rate of the blown air.

In addition, “missing” means that a blow-through is actually generated even if the blow-out prediction score does not reach a predetermined threshold, such as a sudden blow-through due to a sudden increase in the amount of residual iron due to tapping trouble.
These "excessive announcement" and "missing" may or may not occur depending on the setting of the threshold value of the blow-through alarm, and there is a trade-off relationship that if one is suppressed, the other increases. For example, if the alarm issuing condition (threshold of the blow-through alarm) is tightened in order to reduce the occurrence of “excessive announcement”, the occurrence of “missing” increases. It is possible to suppress "excessive announcement" and "missing" by tuning the parameters related to judgment, but this is not a fundamental improvement.

  Therefore, in the operation status evaluation system of the fifth technical example, by extracting transition information until the blast furnace reaches the blow-through state from the data and making it possible to present it to the operator or the operation manager, the phenomenon difference It aims to make it easier to grasp. If such transition information can be expressed explicitly, for example, even if a blow-out prediction alarm is issued, it will be possible to distinguish the difference in the phenomena that have caused it, and it will be possible to take appropriate measures. . For example, even if the blowout prediction score is the same 1.0 (high risk of occurrence), it is possible to judge the excess possibility by referring to the history information and recalling a case that follows a similar history in the past. On the contrary, even if the score is small, it is possible to judge that the risk of blowout cannot be ignored by recalling a case that follows a similar process in the past.

  That is, the operation status evaluation system of the fifth technology example does not evaluate the risk of blow-through by using the index value calculated at a certain time as in the fourth technology example, but the process leading up to the blow-through (change in the situation of the blast furnace ) Is taken into consideration, it is possible to distinguish whether the obtained index value corresponds to a temporary “operational fluctuation” or a transitional tendency (trend) to an unsteady blow-through state. Is intended.

  Specifically, the operating condition evaluation system of the fifth technical example detects instability of the furnace condition such as deterioration of air permeability in the blast furnace, evaluates the risk of risk of blow-through, and determines that the risk is high. If this is done, the system will issue an alarm (attention warning) to the operator. The above-described score is used to determine whether or not this alarm is issued, and when this score exceeds a preset threshold value, an alarm is issued. This alarm is displayed (notified) on a display unit such as a screen, and is similar to the display unit of the first to fourth technical examples.

The operation status evaluation system of the fifth technical example described above is different from the fourth technical example in that not only the judgment regarding the score in the notification of the alarm on the display unit described above but also the judgment regarding the “state transition” are included. That is the point.
In other words, the operating condition evaluation system of the fifth technical example classifies the state inside the furnace into a plurality of categories at each time point, and stores the time during which the state inside the furnace of the classified categories continues. There is. The state transition in the furnace categorized in this manner is the above-mentioned "state transition".

  Then, when the score for determining whether or not the alarm is issued exceeds the threshold, the state transition immediately before that time is selected, and the selected state transition is stored as a pre-stored condition (condition to be satisfied). Collate. As a result of the collation, if the condition is met, an alarm is issued, and if the condition is not met, it is determined not to be issued as "not necessary (excessive alarm)". That is, in the operation status evaluation system of the fifth technical example, the immediately preceding “state” and the duration and the transition thereof, which are stored in the storage means described above, in other words, the immediately preceding “state transition (state transition)” are stored in advance. It is configured to issue an alarm depending on whether the specified conditions are met. For example, the threshold can be set lower than the threshold value of the blow-through prediction alarm, and when the state transition satisfies a predetermined condition, an alarm can be issued to reduce the oversight. Further, even if the threshold value of the predictive alarm is exceeded, if the state transition does not satisfy the predetermined condition, not issuing an alarm can reduce the excessive issuing.

Further, the above-mentioned “category” is for identifying the state related to the state transition, and in the present technology example, the distribution shape of the stave temperature along the height direction of the blast furnace and the pressure fluctuation are used.
For example, when classifying the states in the furnace into categories using the stave temperature distribution in the height direction of the blast furnace, the average and variance in the circumferential direction are calculated for the stave temperature in each height direction of the blast furnace. Then, the calculated average value is "0" and the variance is "1" for normalization (normalization). Next, the average temperature in the circumferential direction of the stave temperature calculated based on the standardized value (normalized average temperature), the maximum variation width of the stave temperature calculated based on the standardized value (maximum Calculate the difference between the value and the minimum value). Further, the coefficient of the first-order term when the stave temperature distribution shape is approximated by a linear function (linear approximation), and the coefficient of the second-order term when the curve approximation is performed by the quadratic function are calculated. Then, using these average temperature, temperature range, first-order coefficient, and second-order coefficient, the state in the furnace is classified by "temperature range", "temperature range", and "shape". In other words, whether the "temperature range" is high, average, or low, or "wide temperature range", the "shape" is the lower temperature protrusion, the upper temperature protrusion, the central temperature protrusion, or the upper and lower temperature protrusion ( Central temperature depression). Then, at each time point, the fitness (affiliation) to these categories is calculated with a real number from 0 to 1.

Moreover, when classifying the state in the furnace into categories using the pressure fluctuations described above, for example, in each pressure gauge arranged in the height direction of the blast furnace, the pressure fluctuations of the furnace body pressure in the last 30 minutes are averaged. Normalize to 0 and variance 1. Then, from the maximum value of the standardized fluctuation value of the furnace body pressure over 30 minutes, the degree of conformity (affiliation degree) to the category of “large fluctuation” is calculated as a real number from 0 to 1.
In the operating condition evaluation system of the fifth technical example, a plurality of categories such as “steady state” and “high temperature variation state” are calculated based on the fitness calculated from the stave temperature distribution and pressure variation in the height direction described above. Based on the combined upper categories (composite categories) and their matching degrees to the lower categories, the matching degree to the upper categories is also calculated.

In the following, first, the “state transition” used in the necessity determination of the alarm issuance in the operation status evaluation system of the fifth technical example will be described.
First, the transition of the stave temperature distribution in the height direction of the blast furnace will be described. For the stave temperature measured from B3 to A6 located in the upper part of the blast furnace from the middle part (berry part) of the blast furnace, the average value μ and the standard deviation σ of the circumferential temperature at each height were calculated and calculated. The mean value and the variance value are normalized to the values of mean 0 and variance (standard deviation) 1 using the frequency shown in the equation (5).

  The thus averaged stave temperature is averaged in the circumferential direction, and the obtained circumferential average temperature is shown in FIG. 26 with the scale of the normalized value as the horizontal axis. The value (normalized value) in FIG. 26 indicates that “0” is the average temperature and “n” is the temperature higher or lower than n × σ from the average temperature. In addition, the vertical axis of FIG. 26 indicates each height position of B3 to A6. In this way, when the normalized average temperature in the circumferential direction at each height is plotted for each height, the state inside the furnace of the blast furnace can be expressed as a polygonal line, and the "stave temperature profile in the height direction of the blast furnace is shown. Can be expressed.

For example, when the state of the blast furnace is a steady state, the stave temperature is near the average temperature at all blast furnace heights, so the line graph extends vertically in a straight line above the value of "0" on the horizontal axis. It becomes the shape of a polygonal line.
Further, in this "stave temperature profile", the shape of this polygonal line also changes with time when the state inside the furnace becomes an unsteady state. Generally, the lower part of the blast furnace (the area where the stave height is B3 to A1 in FIG. 2), the middle part of the blast furnace (the area where the stave height is A2 to A4 in FIG. 2), the upper part of the blast furnace (the stave height is A5 in FIG. 2). If any one of the regions (A6 to A6) protrudes (is biased) to the extremely high temperature side (the right side in the figure), the inside of the blast furnace can be considered to be in an unsteady state.

  The shape of the polygonal line indicating that the inside of the furnace is in an unsteady state is typically as shown in FIG. That is, when the inside of the furnace is in an unsteady state, the lower part of the blast furnace, the middle part of the blast furnace, and the blast furnace are referenced based on the above-mentioned steady state (the shape in which the value of the horizontal axis is "0" is distributed in a straight line). If both the upper part and the upper part and the lower part of the blast furnace protrude to the high temperature side, the shape of the polygonal line also changes. That is, when the temperature inside the blast furnace becomes high, the "line graph (stave temperature profile)" transitions to the high temperature side, that is, the right side of FIG. 26, and if the temperature becomes too high, the range goes out to the right side of the figure. In the past operation cases in which blow-through occurred, the middle part of the blast furnace is often projected or the distribution in which either the height direction of the blast furnace is out of range (the distribution is projected to the right side) is often generated.

For example, FIG. 27 shows a measured stave temperature profile in the actual operation of the blast furnace. Note that the graph existing in FIG. 27 is a line graph of stave temperatures traced back from the present time to the past 8 points at 1-minute intervals in an overlapping manner.
From the left side of FIG. 27, it can be seen that there is a state transition from the lower left graph → the upper right graph → the lower right graph, starting from the upper left graph indicating the steady state. Then, as the state transitions, the stave temperature goes from the lower part to the middle part to the upper part of the blast furnace to the high temperature side, that is, the high temperature part propagates from the lower side to the upper side of the blast furnace and finally outranges and blows through. It can be seen that there is a pattern (scenario) of reaching. In actual operation, there have been many cases in the past where the stave temperature in the lower part of the blast furnace reached a state of a polygonal line, but it was soon restored to a steady state.

  Moreover, as far as the past operation data is seen, when the blow-through alarm is issued when the temperature of the upper part is projected, the blow-through actually occurs frequently. In addition, at the time of start-up after a pause, the temperature in the furnace may fluctuate at high temperature during the 4 to 8 hours before the furnace settles into a steady state. Alarms tend to be over-issued.

  The “state” in the “state transition” described above is defined by the state classification as shown in Table 2. This "state classification" represents in what range (granularity) the continuous measurement values are grouped (classified, clustered), and "the average value of the stave temperature in the height direction", According to the cases of "Stave temperature range" and "Variation of furnace body pressure", they are classified according to different contents (sorting conditions).

  Further, the state classification of the shape of the stave temperature described above is as shown in Table 3 below.

  The "upper protrusion / lower protrusion" shown on the left side of Table 3 indicates the slope (first-order coefficient) of the approximate straight line when the stave temperature profile is approximated by a linear function. In addition, "convex / concave" described on the right side of Table 3 indicates a quadratic coefficient of the approximate curve when the stave temperature profile is approximated by a quadratic function. The sign of the quadratic coefficient when it is approximated by the quadratic function has the opposite sign (minus), but the above-described evaluation of the goodness of fit may be performed by the absolute value of the quadratic coefficient.

The method of calculating the degree of conformity described above relates to the average value of the stave temperature in the height direction of the blast furnace. However, the stave temperature difference (temperature difference of the stave temperature in the height direction of the blast furnace) and the furnace body pressure fluctuation (blast furnace For the fluctuation of the time average of the furnace body pressure over 30 minutes), the degree of compatibility can be calculated by the same procedure.
When the goodness of fit is calculated in this manner, the calculated goodness of fit is used to define an “upper state” as shown in Table 4 below.

    For example, when the state of the actual operation data is evaluated by the "upper state" defined in Table 4, it becomes as shown in Table 5.

For example, in Table 5 described above, when the “data section” is 0.9, the value of “large pressure fluctuation” is 99.6, which indicates that the adaptability of the state of “large pressure fluctuation” is 0.9 or less. It can be seen that the interval of is 99.6% of the whole.
In other words, from Table 5, it is possible to determine the value indicated by the "data section" as the degree of conformity with the actual operation data such as "large pressure fluctuation" and "small stave temperature fluctuation width". For example, when the above-mentioned “large pressure fluctuation” is 0.9 or less, the frequency is 99.6%.

In other words, in actual operation, most of the blast furnace is judged to be in a steady state, and the section where "high temperature fluctuation" or "bottom protrusion" is significant is only about 0.5% of the whole. .
Next, consider extracting the “high temperature fluctuation” or “lower protrusion” section as an unsteady state. Specifically, this unsteady state extraction is performed by "whether the risk of air permeability deterioration or blow-through can be judged based on the increase in stave temperature and increase in fluctuation", or "the increase in residual iron, etc. If the reactor condition deteriorates due to the factor (1), the risk can be judged based on the temperature distribution of the lower protruding type. ”The procedure is as shown in Table 6.

However, the steady-state level and the high-temperature fluctuation level in Table 6 are averages of 30 minutes. For the level of the lower protrusion, the average of 5 minutes is used. This is intended to prevent state subdivision due to subtle variations near the threshold as much as possible.
Generally, at each point in time, a “grade” belongs to multiple states. It is necessary to judge “flexibly” according to the grade whether a certain point of time should be regarded as “steady” or “high temperature fluctuation”. Here, from the viewpoint of data organization, a certain threshold value is set and discrete state setting is performed. The above process is a measure related to such a discrete state determination.

By using such a state, for example,
・ In a certain blow-through case, a state transition from a lower temperature protrusion to a high temperature fluctuation or pressure fluctuation was observed before the alarm was issued.
・ In one case, the temperature drop in the lower part was remarkable, but the blow-through occurred without a high temperature fluctuation or pressure fluctuation.

-In one case, the temperature was fluctuating, but there was no pressure fluctuation and no blow-through alarm was issued.
・ In one case, the temperature fluctuated, but no blow-through occurred.
From the actual blast furnace operation data described above, the blow-through prediction logic as shown in FIG. 28 is summarized.

  That is, as shown in FIG. 28, the blowout prediction logic returns to the initial state when it is determined to be in the “steady state”. If the state is “ELSE”, no transition is made. Furthermore, a transition to any one of "lower temperature protrusion", "high temperature fluctuation", and "pressure fluctuation" is premised on the alarm issuance. Then, when the high temperature fluctuation state continues for a certain period of time or more, a blow-through alarm is issued regardless of the value of the blow-through score.

It was confirmed that the excessive number of official announcements was reduced by actually applying the above-mentioned blowout prediction logic.
"Action effect"
From the above, by adding the above-mentioned state transition to the warning condition of the blow-through alarm, it is possible to suppress the “excessive warning” and “missing” of the blow-through and prevent the productivity reduction of the blast furnace due to unnecessary wind reduction.
"Sixth technology example"
Next, the operating condition evaluation system of the sixth technical example will be described.

  The above-mentioned “blown-out prediction score” of the first technical example is a calculated value at each time point during the operation process of the blast furnace, and always fluctuates. Naturally, when the blow-through prediction is performed using the “blown-out prediction score” having such a large variation, the false determination of the blow-through may increase. That is, it is necessary for accurate prediction of the blow-through to distinguish whether the obtained “open-air forecast score” is due to a temporary change or due to a series of flows (trends) leading to the blow-through.

  For example, a method such as a moving average or a change detection based on a difference from a past trend is applied to the determination of the blowout prediction score (determination whether the value is transient). For example, by linearly regressing past data, the difference between the predicted value of the blowout prediction score predicted from the regression line and the actual measured value of the blowout prediction score is calculated, and the difference between the calculated difference is calculated. A method of determining whether it is a temporary fluctuation or a change (from the viewpoint of whether the difference is larger than expected) is applied.

However, even when the above-mentioned moving average is performed, the retroactive time for calculating the moving average fluctuates according to the operating condition of the blast furnace, and to what extent the past data is used, the fluctuation time (the time interval for calculating the average) ) Decide or change themselves.
Therefore, in the operation status evaluation system of the sixth technical example, the retroactive time for calculating the moving average is determined in accordance with the behavior of the index value. That is, the operating condition evaluation system of the sixth technical example relates to a method of dynamically determining the retroactive time when determining the blow-through prediction score.

  Specifically, the operating condition evaluation system of the sixth technical example determines a retroactive period as a time range in which the retroactive time is changed, and changes the retroactive time within this range to move the blow-through prediction score over the retroactive time. When the average value is calculated, it is determined whether or not the average value (moving average value) of the calculated blowout prediction scores is equal to or higher than a threshold value at which a blowout prediction alarm is issued. Then, if the blowout prediction score for a predetermined retroactive period exceeds the alarm issuance threshold, the alarm is issued or continued, and if it is below the threshold, the alarm issuance or cancellation is performed.

  For example, a retroactive period such as the past 10 minutes, 20 minutes, 30 minutes, etc. is determined, the average value of the blowout prediction scores is calculated while expanding the retroactive time within the range of the retroactive period, and it exceeds the alarm issuing threshold value. In this case, if the alarm has already been issued, the alarm is issued (the current value of the blow-through prediction score may be less than or equal to the threshold). If the alarm has not been issued yet, the alarm is issued.

On the contrary, if the blow-through score prediction value is less than or equal to the threshold value at any retroactive time within the retroactive period, the alarm issuance is canceled if the alarm is already issued, and the alarm not-issued state is continued if the alarm is not yet issued.
Actually, as shown in the flowchart of FIG. 29, the flow is branched and different processing is performed depending on whether or not the flag for issuing the blow-through alarm is already set. When the flag for issuing a blow-through alarm is not set, in other words, when a new blow-through alarm is issued, an alarm is issued immediately when the instantaneous value of the blow-through prediction score exceeds the threshold for issuing an alarm. To do.

When the flag for issuing the blow-through alarm is already set, if the blow-through prediction score is equal to or higher than the threshold value for issuing the alarm, the alarm is continued to be issued even if the immediate value is equal to or lower than the threshold value, and a more careful cancellation decision is made.
Note that the operation status evaluation system according to the present embodiment sets the retroactive time to 0 when the alarm is issued from the state where the alarm is not issued to the issuance. That is, when the instantaneous value of the blowout prediction score exceeds the threshold value for alarm issuance, an alarm is immediately issued.

As shown in FIG. 30, even when the alarm is issued, since the moving average is performed at 1-minute intervals until the retroactive period elapses, even if the blow-through prediction score falls below the threshold, the moving average is calculated. The alarm will not be released until the result is below the threshold. For example, when the retroactive time is set to a maximum of 10 minutes, the alarm is released with a delay of 10 minutes. On the other hand, at the time of cancellation, the warrant is maintained for a maximum of 60 minutes, but in reality it is maintained for 20 to 30 minutes due to the sharp drop in the index value. As a result, the alarm issuance can be maintained for a temporary decrease after the elapsed time of 65 minutes.
"Seventh technology example"
Next, the operating condition evaluation system of the seventh technical example will be described.

The operation status evaluation system of the seventh technical example extracts transition pattern information to a non-steady state from time series information of a plurality of sensors and uses it for state diagnosis and risk prediction, and at the same time, designates a transition pattern and then It is possible to search and extract a time series section according to the above.
Specifically, the operation status evaluation system of the seventh technical example has the following functions (1) to (3).
(1) State transition sequence extraction function The state transition sequence extraction function is defined in advance such as "higher than usual" or "larger than usual" with respect to sensor measurement items such as temperature and pressure and values acquired in time series. It is determined that it is established based on the function of calculating the degree of conformity with the unsteady state (hereinafter simply referred to as “state”) and the function of storing the threshold value for determining whether or not the state is established, and the measurement value at each time point. The function to obtain and store the time series information of the state set, that is, the established state set, and the difference between the state set at the preceding time point and the state set at this time point at each time point, that is, the newly established state and the non-established state A function to obtain a set of states and selectively store the time when a newly established or not established state set is not an empty set as a "state change point" and the state for all past time points. Change point And a function of extracting a state transition sequence in which only the changed state set is extracted.
(2) Monotonically increasing sequence extracting function The monotonically increasing sequence extracting function is a set in which the state set in the above state transition sequence is monotonically increasing, that is, the subsequent state set is a set including all the immediately preceding state sets and further adding new states thereto. This is to extract a certain partial series.
(3) State transition pattern extraction function The state transition pattern extraction function is included in the state set as the interest information of the state transition sequence with respect to the above-mentioned monotonically increasing sequence. Number of states,
-Presence / absence of inclusion of specific state-Start time and end time of state series (at the time of the first and last state change)
When the conditions for the above items are set, the state transition series that matches the conditions is selected as an object of interest, and the state transition occurrence period is extracted.

Here, the "state (unsteady state)" corresponds to a state in which the measurement value deviates from the normal value range or an abnormal state occurs. From the viewpoint of process stability, it is preferable that such a condition is not established. On the other hand, it may be desirable to issue an alarm (caution, warning) to the operator to establish the condition.
According to the present invention, scenarios related to the propagation of abnormalities and their frequencies are extracted. Extracting such a scenario is useful in the following two points.

・ It becomes possible to classify and frequency of causes and phenomena based on patterns of propagation or spread of abnormalities. In general, the non-steady state often includes various factors and phenomena (“all but normal” is a non-steady state), and such classification is useful for identifying the cause and planning a countermeasure.
-If you are currently in a non-steady state, it is possible to search for similar scenarios including the background, it is possible to search future predictions based on past cases and past correspondence cases, which is useful for skill transfer. .

For example, Japanese Patent No. 5048625 “Abnormality Detection Method and System” discloses a change detection method for detecting an abnormal event from a change in the statistical properties of time series information. However, the multidimensional data is arranged according to time, and it is difficult to use it for process problems such as blast furnace in which influences between data items, differences in interference degree, and time delay degree depend on the situation. (Applicable to mechanical systems, processes that do not need to consider time delays and disturbance fluctuations)
In addition, in the "action prediction method and action prediction system" of Japanese Patent No. 5199152, a human is equipped with a position and action measurement sensor such as a GPS sensor, and from the measured value, a "scene" such as commuting or taking a walk. , And the method of constructing the behavior model by the transition is shown. In the present invention, a combination of individual states defined in each data item, that is, a state set is regarded as one "process state" and its change is followed. That is, the combination method of the individual states is the defining element of the process state.

On the other hand, this "action prediction method and action prediction system" defines the state based on the value of a specific sensor, and cannot be applied when it is not clear which data item contributes to the state.
That is, the operation status evaluation system of the seventh technical example has an effect that it can be applied when a physical phenomenon or a state definition is clear or intentional.

Next, the operation status evaluation system of the seventh technical example will be described with a specific example.
The measurement items currently measured are {P1, P2, ...}. These are, for example, temperature, pressure, parameter values characterizing the temperature distribution shape, and the like. Although it may be an index value calculated using the measurement value of the sensor, it will be described as “measurement value” below.
Next, the unsteady state is defined for each measured value. For example, assuming that the long-term average of the measured values of P1 (for example, the average of the latest one month) is μ and the standard deviation is σ, when the measured value exceeds μ + 3σ, the state S1 of “increase in value of P1” is established. Alternatively, for the measured value of P2, a value obtained by dividing (dispersion of the last 30 minutes) by (dispersion of the latest 120 minutes) is calculated, and when this value exceeds a certain value N, a state S2 of "variation of P2" Is established. Let S = {S1, S2 ,. ． ． }, The success or failure of the state is determined from the measured value at each time. A set of established states is represented by A1, A2, ..., And | A1 | is the number of elements, that is, the number of established states.

As an example of the state transition sequence, (empty set, time t0) ⇒ (state set A1, time t1) ⇒ (state set A2, time t2) ⇒ (state set A3, time t3) ⇒ (state set A4, time t4) => (State set A5, time t5) => (state set A6, time t6) => (empty set, time t0).
The number of elements in the set is set as | empty set | = 0, | A1 | = 1, | A2 | = 3, | A3 | = 4, | A4 | = 3, | A5 | = 4, | A6 | = 1. To do.

Note that only the time when the state set has changed is extracted. For example, from time t1 to t2, the state included in A1 continues to be established.
Here, the subsequences in which the number of states does not decrease are A1⇒A2, A1⇒A2⇒A3, A2⇒A3, A4⇒A5.
In the frequency calculation, the number of times these sequences (three in length 2 and one in length 3) appear in the entire state sequence is calculated.

Although the state continuation time (eg t2-t1 etc.) is ignored here, it is possible to add a time condition such as using only those whose minimum duration of each state exceeds α as an "effective series". .
-The operating operator specifies a state sequence (for example, "A1⇒ * ⇒A3" (state transition sequence pattern from A1 to A3 after passing through some state) and obtains a time interval corresponding to it. For example, A1 indicates "part X If A3 is "Emergency stop of device X" in "Temperature rise at the same time", a period that has undergone such a process is extracted, and by investigating the situation and actions during that period, an unsteady state of emergency stop Useful information can be obtained to reduce the overall risk.

Alternatively, when the “temperature rise at the part X” is currently established, a risk scenario leading to “emergency stop of the device X” can be searched and a method of preventing it can be considered.
Such similar scenario search is useful for skill transfer or experience sharing.
For example,
・ Stave temperature of large scale S4 to S6 becomes high ・ Fusion zone behavior It is estimated that the root position of cohesion zone rises from stave temperature ・ Deviation large The stave temperature in a specific direction around the zone is high (peripheral flow is occurring)
・ Blower pressure is higher than usual. ・ Burning pressure is higher than normal. ・ Top gas temperature is higher than normal. ・ Alarm red Monotonic state transition extracted from actual case when the condition is defined as 0.8 or more. The sequence is illustrated in FIG. Looking at the conditions before and after the occurrence of the blow-through, it is considered that the "furnace top size" is an effect of the blow-through. It can be seen that the states gradually pile up (unsteady state becomes noticeable), leading to blow through. Also, it can be seen that the number of established states is effective as a measure of unsteady state.

In addition, if a state transition model is obtained, time-series interval search based on state transition can also be performed.
For example, search condition: {large scale} → * → {large blast pressure, *} ({large scale} from a single state to a state in which a large blast pressure is added to a state after an intermediate state including "large scale") Given the condition of (time series section), two sections are obtained from FIG. However, in this case, not only the monotonically increasing series but also a series of time series until “large scale” and “large blast pressure” are added are extracted.

Here, the process of peripheral fluidization has been completed. Therefore, if the extraction conditions are set as {large scale, *} → {large drift, *} → {large blast pressure, *}, two different sections can be similarly obtained.
In this way, by calling a pattern on the state transition sequence (conditions that the sequence must satisfy) a “scenario”, it is possible to specify a scenario and extract a case. The series of system configurations described above is shown in FIG.

As described above, by using the first to seventh technical examples, not only the risk of occurrence of blow-through at the time of operating the blast furnace, but also the scale when the blow-through occurs, and the prediction By displaying the result to the operator, it becomes possible to more reliably prevent blow-through and to perform stable blast furnace operation.
As described above, the embodiments disclosed this time are to be considered as illustrative in all points and not restrictive. The scope of the present invention is shown not by the above description but by the scope of the claims, and is intended to include meanings equivalent to the scope of the claims and all modifications within the scope.

  For example, although the prediction monitor 11 that visually appeals to the operator M is illustrated as a display device that presents various information to the operator M, a device that transmits information to the operator M by voice may be used as the display device.

1 Blast furnace 2 Iron skin 3 Refractories 4 Tuyers 5 Raceway 10 Prediction system 11 Prediction monitor (display)
12 Computer 13 Predictor M Operator S Pig iron

Claims (5)

A data storage unit that stores, as time-series data, a value that is periodically measured by a sensor attached to the blast furnace and a value calculated from the measured value.
A prediction unit that performs a prediction regarding blow-by using the value stored in the data storage unit,
A value stored in the data storage unit and a display unit for displaying the prediction result predicted by the prediction unit so that the operator can confirm the result;
Have
The prediction unit is
A data acquisition unit for acquiring time series data from the blast furnace,
Existence in the time-series data, the fluctuation of the pressure around the blast furnace and in the height direction, and the index value representing the ventilation state using the furnace wall temperature is obtained, and the furnace body pressure, the blast pressure, the furnace top gas temperature , and obtains a score indicating a poor furnace situation with staves temperature, the analyzing unit which calculates a risk of blow by using the score and the index value,
Have a,
The analysis unit has a scale index value indicating the degree of heat loss when blow-through occurs using the stave temperature in the time-series data,
Operating conditions characterized in that the temperature of the upper part of the blast furnace and the temperature distribution in the height direction in the upper part of the blast furnace are configured so that the scale index value becomes large when they become abnormal values. Evaluation system. The display unit is
Equipped with a means for displaying an alarm issuance at the time when the risk of blowout exceeds a predetermined threshold value,
While the average value calculated from the index indicating the risk of blow-through calculated over the past period is higher than a predetermined value, the alarm display indicating the risk of blow-through is continued,
The operation condition evaluation system according to claim 1, wherein when the alarm is issued and the furnace body pressure exceeds a threshold value, a wind reduction instruction is given. The prediction unit is
For each of the fluctuation of the furnace body pressure, the deviation of the stave temperature in the circumferential direction, and the change in the temperature distribution of the stave temperature in the height direction, an index that becomes a large value when deviating from the normal value is set, and these 3 The operating condition evaluation system according to claim 1, wherein a minimum value of the two indexes is set as an index value that represents a poor ventilation condition. The prediction unit is
That heat is accumulated in the upper part of the blast furnace, the position of the root of the cohesive zone in the blast furnace is higher than the normal position, and a higher temperature than usual is confirmed with a sensor of a specific height of the blast furnace. , The blast pressure is higher than normal, the furnace top temperature is higher than normal, each is defined as a precursory state of blow-through, and the presence or absence of these five precursory conditions is calculated as a score. The operating condition evaluation system according to claim 1 or 2, wherein The operating condition evaluation system according to claim 2 , wherein an index obtained by combining the index value, the score, and the scale index value is calculated as a blow-through risk index.