KR19990052506A - Pressure control method and device in hood of steelmaking converter - Google Patents

Abstract

The present invention relates to a control method and apparatus for maintaining a constant pressure in a hood located at the top of a converter in an iron making operation in a steel mill, and a change in the opening degree of a damper in a method of adjusting the opening degree of a damper to control the pressure in the hood. The invention has been invented to solve the problem of the conventional method, which is unable to efficiently control the pressure in the hood due to the delayed time required to propagate to the pressure in the hood.

According to the present invention, the pressure control in the hood is performed by predicting the pressure value in the hood after the time delay using data on damper opening ratio, skirt height, flame amount, and the like, which determine the dynamic characteristics of the hood pressure gauge. It is done.

According to the pressure control method and apparatus in the hood of the present invention, by keeping the pressure constant by reducing the change in the pressure in the hood to stabilize the steelmaking operation using the oxygen-fusing steelmaking converter, high-purity CO gas recovery rate It can reduce the production cost by improving the cost, and also provides the effect of reducing the air pollution by suppressing the external emissions of CO gas.

Description

Pressure control method and device in hood of steelmaking converter

The present invention relates to a hood pressure control method and apparatus for maintaining a constant internal pressure of a hood located at an upper part of an converter in an oxygen smelting steelmaking process of a steel mill, and in particular, the hood pressure predicting means for predicting the hood pressure and the hood pressure prediction means. A pressure control method and apparatus in a hood of a steelmaking converter adapted to control the hood pressure by utilizing an output from the means.

Fig. 1 shows the construction of a facility for carrying out steelmaking using the oxygen blowing steelmaking method, which is one of the major steelmaking methods in the world, which uses molten iron in the converter 1 through the oxygen lance 3. 2) It is a facility that oxidizes and removes the molten impurities in molten iron by spraying them on the surface. The exhaust gas in the converter generated by the oxidation reaction is cooled while passing through the primary dust collector 7 through the duct 6 and dust in the exhaust gas is removed. The exhaust gas which passed through the primary dust collector 7 is recooled and collected in the secondary dust collector 9 and then exits through the manned blower 10. The collected exhaust gas is sent to the stack 12 through the watertight side 11 to be burned and then blown into the atmosphere, or when the CO gas purity in the exhaust gas is high, it is stored in the storage tank 13 for reuse as fuel. . The exhaust gas is stored in the storage tank 13 only when the CO gas purity is high, and in order to increase the exhaust gas, there should be no air flowing into the gap between the skirt 4 and the converter 1, and for this purpose, the internal pressure of the hood is appropriately adjusted. You must do it. If the pressure inside the hood is not properly adjusted and the hood pressure is too low, outside air will flow between the inlet 1 and the skirt 4, causing the outside air and CO gas to secondary burn in the duct 6 to explode. Of course, the CO ₂ is regenerated to reduce high purity CO gas recovery rate, and if the pressure inside the hood is too high, the exhaust gas in the converter is leaked to the outside and a large amount of flame is leaked, or the molten iron in the converter overflows out of the converter. do.

In order to prevent such phenomena, as shown in FIG. 1, a hood pressure controller which measures the pressure in the hood through the pressure measuring sensor 14 and the hood pressure measurement value and the reference hood pressure value adjust the damper 8. The hood pressure controller 15 adjusts the opening degree of the damper 8 according to the difference between the hood pressure measurement value from the pressure measuring sensor 14 and the reference hood pressure value. It was. Here, since the suction force of the attracting blower 10 is fixed, the opening degree of the damper 8 becomes the only means for determining the amount of exhaust gas drawn through the duct 6. The hood pressure controller 15 recognizes the difference between the current hood pressure measurement value and the reference hood pressure value and reduces the difference between the venturi damper located between the primary dust collector 7 and the secondary dust collector 9. (8) Calculate the opening rate. In the hydraulic actuator 16, the opening degree of the damper 8 is adjusted in accordance with the calculated opening degree signal. The hood pressure controller 15 opens the damper when the pressure in the hood is higher than the reference pressure value to increase the amount of exhaust gas sucked, thereby lowering the pressure in the hood, and conversely, when the pressure in the hood is lower than the reference pressure value. It is operated to increase the pressure in the hood by closing the amount of exhaust gas sucked in.

2 is a block diagram of a conventional hood pressure control system using the hood pressure controller 15 as described above. P ( _γ ) S, which is the reference hood pressure signal 17, and E (s), which is an error 19 of P (s), which is the current hood pressure value 28 measured by the hood pressure measuring sensor 14, are obtained. The transfer function 20 G _c (s) of the hood pressure controller for calculating the opening ratio of the damper from the first retarder 18 and the error 19 E (s) that is the output value of the first recoil 18 is have. Here, the transfer function 20 G _c (s) of the hood pressure controller 15 showing the relationship between the damper and hood pressure is expressed by the following equation (1). In Eq. (1), K _p is proportional gain, K _d is differential gain and K _i is integral gain.

D (s), which is the output value of the transfer function G _c (s) of the hood pressure controller 15, means the opening ratio 21 of the damper. The transfer function (24) G _p (s) between the damper opening and the hood pressure as the input of D (s), which is the damping rate (21) of the damper, is the transfer function (22) G _d (s) of the duct section and the skirt section. It consists of the transfer function 23 G _s (s). The G _d (s) and G _s (s) are as shown in the following formulas (2) and (3), respectively.

In Equation (2), K is the transfer gain of the duct 6, T _d is the delay time due to the duct length between the damper 8 and the hood pressure sensor 14, T is the time constant of the duct, respectively do. In Equation (3), R is a gain of the skirt determined according to the height of the skirt, and is in inverse proportion to the height of the skirt. Denotes the time constant of the skirt portion, which is proportional to the volume V _o of the hood portion and inversely proportional to the current hood pressure value P _o .

As described above, the transfer function G _p (s) representing the relationship between the damper and hood pressure is a time delay element. The nonlinear characteristic is included, and when the controller according to the transfer function G _c (s) as shown in Equation (1) is used for the system of such a nonlinear characteristic, after the time of disturbance 36 as shown in FIG. The furnace hood pressure value 35 changes with a large fluctuation range. The reason why the hood pressure value changes with a large fluctuation range without converging with the reference pressure value is because of the time delay element, which is a nonlinear characteristic described above. e ^{-T d} Is in the transfer function G _p (s). In order to compensate for this time delay factor, in control engineering theory, instead of feeding back the current hood pressure value 28 (P (s) measured as shown in FIG. 3), the time delay factor at G _p (s). A method of feeding back the output 33 of the predictive model 32 G ^* (s) Z (s) which has been extracted only is given. Here, Z (s) means the predicted value after T _d of the hood pressure.

As described above, the time delay element at G _p (s) instead of feeding back the measured current hood pressure value 28 P (s). The method of feeding back the output 33 of the predicted model 32 G ^* (s) Z (s) is called Smith predictive control technique. The hood pressure control result using this method is shown in FIG. Comparing the hood pressure 37 in FIG. 5 and the hood pressure 35 in FIG. 4, it can be seen that the fluctuation range of the hood pressure is significantly reduced after the timing 36 when disturbance occurs.

However, in reality, the various gains included in the above equations are impossible to obtain and an alternative method is required to solve them.

An object of the present invention is to implement the predictive model 32 G ^* (s) as described below in order to solve the problems of the conventional hood pressure control method described above, that is, a skirt for determining the transfer function of the hood pressure gauge Provides a hood pressure predicting means for predicting the hood pressure after the time delay described above using data on the height, the flame quantity of the furnace section, the current hood pressure value, and the current damper opening degree, and the hood which is an output value of the hood pressure predicting means. It is an object of the present invention to provide a method and apparatus for improving hood pressure control by compensating a time delay element in a duct by feeding back a pressure predictive value and adjusting a damper.

1 is a schematic view of the oxygen blowing steelmaking process related to the present invention,

2 is a block diagram of a conventional hood pressure control system;

3 is a block diagram of a hood pressure control system using Smith prediction technique;

4 is a control result diagram of a conventional hood pressure control system;

5 is a control result diagram of a hood pressure control system using Smith prediction technique;

6 is a hood pressure control block diagram using a fuzzy neural network in the present invention,

7 is a hood pressure control result using the present invention.

Explanation of symbols on the main parts of the drawings

1: steelmaking converter 2: molten iron 3: oxygen lance

4: skirt 5: hood 6: duct

7: primary dust collector 8: damper 9: secondary dust collector

10: manned blower 11: waterproof side 12: stack

13 exhaust gas storage tank 14 hood pressure sensor 15 hood pressure controller

16: hydraulic actuator for damper adjustment 17: reference hood pressure value

18: first acceleration and winding 19: error signal

20: transfer function of hood pressure controller 21: damper control signal

22: duct transfer function 23: skirt transfer function

24: Transfer function of hood pressure gauge 25: Signal of inflow amount of exhaust gas in duct

26: external air inflow signal 27: second acceleration and rewind

28: Hood pressure measurement value signal 29: First acceleration and rewind

30: transfer function of hood pressure controller 31: transfer function of hood pressure gauge

32: predictive model of hood pressure gauge 33: hood pressure prediction value signal

34 error 35 Hood pressure signal 36 Disturbance application test

37: hood pressure signal 38: hood pressure prediction means 39: learning means

40: Prediction error calculation block 41: Time forward circuit

42: Hood pressure predicted value signal 43: Actual hood pressure value signal after time delay

As a technical means for achieving the above object, the present invention comprises the steps of inputting data on the skirt height and the flame amount of the furnace section, the current hood pressure value and the opening degree of the damper to determine the transfer function of the hood pressure gauge; And identifying the difference between the predicted value and the actual value for the hood pressure based on the data, and learning the network of the hood pressure predicting means in a direction to minimize the difference between the predicted value and the actual value. Provided is a pressure control method and an apparatus for implementing the same.

Hereinafter, with reference to the accompanying drawings of the present invention will be described in more detail.

Fig. 6 shows the construction of the apparatus according to the present invention, which includes a hood pressure predicting means 38 consisting of a fuzzy neural network and learning means for learning the hood pressure predicting means 38 by the following rules. (39). The hood pressure prediction means 38 x ₁ , x ₂ , x _3, x _4, x _5, x ₆ Receive data marked as x ₁ Silver flame, x ₂ The skirt height, x ₃ Opening degree of the damper, x ₄ Is the rate of change of damper opening, X ₅ is the current hood pressure, and X ₆ is the rate of change of hood pressure. The hood pressure predicting means 38 calculates a transfer function represented by the following equation (4), (5), (6) and (7).

A _ij = e- ^{[w ik (x i -w k )] 2} ---------(4)

In Equation (4), A _ij means an output value of the j th fuzzy membership function for the i th input. Here, the j th membership function is a normal distribution curve with a center value of w _jc and a standard deviation of w _jg .

Μ _k in Equation (5) means the goodness of fit of the k th fuzzy rule Denotes the normalized goodness of fit of the k th fuzzy rule. In equation (6), f _k (·) is the normalized goodness of fit of the k th fuzzy rule Is the weight of and w _ik is the weight of the k th rule for the I th input.

In Equation (7), N is the total number of fuzzy rules, (t + T _d ) is the estimated hood pressure value after the delay time T _d seconds described above.

The learning means 39 calculates a transfer function such as Equation (8), Equation (9), and Equation (10).

E commonly used in the said Formula (8), Formula (9), and Formula (10) is calculated | required by following Formula (11). T _s is a sampling period which is a time interval for reading hood pressure from the hood pressure sensor, and η means a learning rate.

In the formula (11), P (t + T _d ) means the actual hood pressure value after the delay time, (t + T _d ) means a hood pressure predictive value which is an output of the hood pressure predicting means 38.

The time leading circuit 41 receives the current hood pressure signal X _{5 and} sends the actual hood pressure signal 43 after a time delay to the predictive error calculating block 40, and the predicted error calculating block 40 receives the hood pressure. The error is calculated between the hood pressure predicted value signal 42 from the predicting means 38 and the actual hood pressure signal 43 from the time leading circuit 41, and the signal is sent to the learning means 39. 39), based on this, to calculate the transfer function of the above-described equations (8), (9), (10) to learn the hood pressure prediction means 38 consisting of a fuzzy neural network in a direction to minimize the error, this process As the repetition is repeated, the error between the hood pressure predicting means 38 from the hood pressure predicting means 38 and the actual hood pressure value is minimized, so that the pressure in the hood can be kept more constant.

7 shows the results of the hood pressure prediction using the hood pressure prediction means 38 and the learning means 39 made of the above-described fuzzy neural network. Comparing the actual hood pressure signal 44 and the hood pressure prediction signal 45 shown here, it can be seen that the hood pressure prediction signal 45 is predicted by a time delay of 2.7 seconds.

As described above, according to the method and apparatus for in-hood pressure control of a steelmaking converter according to the present invention, it is possible to predict the hood pressure after a delay time required for the opening and closing of a damper for performing hood pressure adjustment to be transmitted to the pressure in the hood. By applying this value to the damper controller, it is possible to reduce the fluctuation of the hood pressure, which is a problem due to the delay time, and thus to keep the pressure in the hood constant, thereby further stabilizing the steelmaking operation using the oxygen blown steelmaking converter. In addition, it is possible to improve the recovery rate of high purity CO gas with high calorific value, thus reducing the production cost and reducing the air pollution by suppressing the external emission of CO gas.

Claims (3)

As a method for maintaining a constant pressure in the hood located at the upper portion of the steelmaking converter in the steelmaking process of the oxygen blowing steelmaking industry, Inputting data on the height of the hood skirt determining the transfer function of the hood pressure gauge and the flame volume between the furnace section, the current hood pressure value, and the opening degree of the damper for adjusting the pressure in the hood; Estimating the hood pressure after time, and learning a network for performing the hood pressure prediction step in a manner of identifying a difference between the predicted hood pressure value and the actual hood pressure value and minimizing the difference. A pressure control method in a hood of a steelmaking converter characterized in that the. The method of claim 1, wherein the step of estimating the hood pressure is performed through a hood pressure predicting means comprising a fuzzy neural network, wherein the hood pressure predicting means minimizes the difference between the estimated hood pressure value and the actual hood pressure value. A pressure control method in a hood of a steelmaking converter characterized in that it is learned by a rule. As a device for maintaining a constant pressure in the hood located on the upper part of the steelmaking converter in the steelmaking process of the oxygen blowing steelmaking industry, Data on the amount of flame, the skirt height, the opening degree of the damper 8, the rate of change of the damper opening, the current hood pressure, and the rate of change of the hood pressure ( x ₁ , x ₂ , x _3, x _4, x _5, x ₆ Hood pressure predicting means (38) consisting of a fuzzy neural network for generating hood pressure predictive value signal (42) on the basis of the input and input data, and hood pressure predicting value signal and time delay from the hood pressure predicting means (38). A predictive error calculation block 40 for calculating an error between the predicted value and the actual value of the hood pressure by receiving the actual actual hood pressure signal, and the hood pressure predicted value and the actual hood pressure based on the signal from the predicted error calculation block 40. And a learning means (39) for learning the hood pressure predicting means (38) in a direction of minimizing the difference between the values.