JPH03160213A - Automatic burner for shaft furnace, fuzzy theory applied thereto - Google Patents

Abstract

PURPOSE:To save manpower in the operation of a furnace and stabilizing the quality of a product based on the stable amount of melting and the stable melting temperature of copper by controlling the melting depending upon a retaining amount and its variations with time. CONSTITUTION:Three or more membership functions from NB(negative big) to PB (positive big) are set at a fuzzy inference section with respect to the deviation E between a desired retaining amount and an actual retaining amount and also the variation DELTAE of the deviation with time, and the variation amount of a combustion level at the time of respective combination is determined. A new combustion level is set by integrating the variation amount determined by the fuzzy inference to the combustion level under actual condition. The combustion level preset by a fold-line function on the basis of the combustion level determined by the fuzzy inference section is converted into an air pressure in each zone according to the relation with the air pressure in each zone. Then, a coefficient that varies stepwise with respect to the melting temperature of copper is set for each zone, and this coefficient is multiplied by the air pressure in each zone, which has been found previously, to make a correction and to set a final burner air pressure.

Description

[Detailed description of the invention] [Industrial application field]

This device relates to a method for controlling the combustion amount of an LNG burner in a shaft furnace.

[Conventional technology]

Conventionally, in the combustion control of shaft furnaces, the retained amount is divided into seven levels, the combustion amount corresponding to each level is set, and the combustion amount is determined only by the size of the retained amount. Conventional shaft furnace control will be explained in more detail with reference to FIG. The shaft furnace 1 is a device that inputs steel material, heats it, melts it, and taps it. In the shaft furnace 1, LNG burners are installed in the A zone, B zone, and C zone from the bottom. A gas supply device is used that mixes fuel gas while maintaining a constant l-containing ratio according to changes in burner air pressure.By adjusting burner air pressure,
You can adjust the firepower of Pana. The copper melted in the shaft furnace 1 is conveyed to the holding furnace 2. The holding furnace 2 is a circular container, and has a tap hole at the end face of the cylinder. The holding furnace 2 is supported so as to be rotatable around the cylindrical axis. The rotation angle of the cylinder around the axis is called the tilt angle θ. Molten copper is taken out from the tap and processed into a desired shape through casting, rolling, and other processes. The height of the tap hole corresponds to the level of the molten copper in the vessel. Therefore, there is a unique relationship between the amount H of copper held inside the holding furnace (the amount held) and the tilt angle θ. The smaller the tilt angle θ is, the larger the holding amount H is. If θ is large, the holding JIH is small. This depends on the geometry of the holding furnace. Actually, as shown in graph (A), the relationship between the tilting angle θ and the holding amount H is determined in advance using a polygonal line function. The holding amount H is determined by knowing the tilting angle θ. This is the tilt angle retention amount conversion. The amount of molten copper leaving the holding furnace 2 is controlled to be approximately constant. Therefore, the fact that the holding amount H in the holding furnace 2 is large means that the amount of steel melted and tapped out in the shaft furnace 1 is large. On the other hand, the fact that the retained ffiH is small means that shaft furnace 1
This means that the rate at which the steel material melts is small. In reality, it is desirable that the amount retained be constant. When this value fluctuates, the quality of the copper products produced in subsequent processes tends to become inconsistent. Therefore, a target retention amount is set, and the air pressures of the burners A, B, and C of the shaft furnace are adjusted so that the actual retention amount H approaches this value. Conventionally, control has focused only on the retained amount H. In other words, if H is small, the thermal power of the burner of the shaft furnace 1 is increased to increase the melting of copper and increase the output amount. On the other hand, if H is large, the thermal power of the burner of the shaft furnace 1 is reduced to suppress the melting of copper. Then, the retained amount and burner air pressure were converted as shown in graph (a). In reality, the burner air pressure is continuously changed in a step-like manner.In this case, in the form of a step function with 7 steps, the retention amount●The 0 burner that was converting the burner air pressure is Since there are three burners, B and C, specify the air pressure of the three burners depending on the amount of burner to be held. The maximum value of the holding amount H is 15t in this example, but for example, when H is 8 to 9t, the burner air pressure is 800mmAq for A and 900mmAq for B.
Specify that XC is 950 mmAq. Such air pressure designation is made at regular sampling times. The sampling number is represented by k here. ■ θ, QA, etc. indicate the holding amount, tilting angle, and air pressure of the A burner at the k-th sampling time. The same thing is repeated, but QAk1 at time k
QBk1QCk is determined by the held amount H2 at that time, and does not require any values at previous times. The burner air pressure values of A and B1C thus determined are output to the indicating controller 3 and the position controller 4. The output of the positioner becomes a valve operation signal 7 from the changeover switch 5, which rotates the motor 6 in the forward or reverse direction to adjust the opening degree of the valve 8. The rotation angle of the motor 6 becomes a signal representing the valve opening degree IO, and is returned to the position indicator 4. A feedback system that outputs a signal from the positioner 4 and returns it again sets the opening degree of the valve so that the desired air pressure is achieved. In this way, the firepower of burner A in zone A is adjusted. The same thing is done simultaneously for burners B and C in the B zone and C zone.

[Problem to be solved by the invention]

In conventional combustion control, control was performed only based on the size of the retained amount, so for example, if the current value is larger than the target value, even when the retained amount is moving closer to the target value, it will move away from the target value. Since the combustion amount was exactly the same even when the fuel was changing, there was a large variation in the amount retained. Furthermore, there is no control mechanism based on the molten copper temperature, and it is necessary to manually adjust the amount of combustion in order to stabilize the molten copper temperature.

[Means to solve the problem]

The shaft furnace automatic combustion device of the present invention includes a shaft furnace that melts copper and has a plurality of burners A1B, . . . , and a holding furnace that holds molten copper tapped from the shaft furnace. This is a device that automatically controls the combustion amount of the burner of a shaft furnace for the purpose of stabilizing the amount and temperature of the molten copper.It uses the holding furnace tilting angle and the molten copper temperature as input signals, and controls the electric valve of the burner of the shaft furnace. It uses a valve operation signal as an output signal, and includes a preprocessor section, a fuzzy inference section, a postprocessor, and a hardware section. It calculates the holding amount from the tilt angle according to the function, calculates the deviation E between the preset target holding amount and the current holding amount, and calculates the change ΔE in the deviation E in the preset control cycle. The fuzzy inference section calculates PB (positively large) to NB (negatively large) for E and ΔE.
Set a membership function of 3 or more to The amount of change in combustion level Δ for each combination of and ΔE
Set a rule map that gives During inference, the relationship between the retained amount and combustion level at the start of control is set in advance as a line function, the initial value of the combustion level is determined according to that relationship, and the combustion level is determined by multiplying this initial value by the inference result. The post processor section sets the relationship between the i@burning level and the burner air pressure of each zone in advance using a line function, and calculates the air pressure of each zone from the combustion level determined by the fuzzy inference section according to the line function. Pressure PAN PB. Calculate the coefficient that changes stepwise depending on the temperature of the melt #I as A1B1C
A coefficient AKNBK that is set in advance for each zone and should be applied to each zone from the current water temperature.
, and the coefficients AK, BK. obtained in the previous process. ．． ．． is the air pressure of each zone PAN P[]-. ．． The air pressure of each zone obtained by this is output to the indicating controller as the final control result. The present invention is characterized in that it outputs a valve operation signal of an electric valve so that the valve operation signal is as follows. To put it more simply, it is as follows. ■This device uses a fuzzy inference unit to calculate three or more member sysob functions from NB (negatively large) to PB (positively large) for the deviation E between the target retention amount and the actual retention amount and its time change ΔE. The amount of change in combustion level for each combination is determined in advance. ■The amount of change determined by fuzzy inference is added to the current combustion level to obtain a new combustion level. - Based on the combustion level determined by the fuzzy inference section, the combustion level set in advance using a polygonal line function is converted into each zone air pressure according to the relationship with each zone air pressure. (2) Next, a coefficient that changes stepwise with respect to the molten copper temperature is set for each zone, and this coefficient is applied to the previously determined air pressure of each zone to correct it and obtain the final burner air pressure.

[For production]

(1) Since this device performs control using the retained amount and changes over time in the retained amount, there is less variation in the retained amount and it is easier to stabilize at the target retained amount compared to conventional control using only the retained amount. ■In the combustion control of shaft furnaces, there is a long delay from when the burner combustion amount is controlled until changes in the amount of melted copper and molten copper temperature appear, and the delay time fluctuates depending on the condition of the material in the furnace. Because it is difficult to formulate formulas and model, PID
It was difficult to apply conventional control methods such as control. (3) In this fuzzy inference, it is possible to automatically control the burner combustion amount by expressing the qualitative grading of the retained amount and its temporal change using a membership function, and by creating rules for actual operator work. became.

【Example】

In FIG. 1, the control device of the present invention is shown in the middle. Molten copper comes out of the shaft furnace 1 and enters the holding furnace 2. There is a unique relationship between the tilting angle θ of the holding furnace 2 and the holding amount H, and H can be determined from θ. The holding furnace 2 is tilted and molten copper is discharged from the tap and used for processes such as casting and rolling. Since this speed is constant, a large holding ffiH means that the molten copper action in the shaft furnace 1 is active. In this case, it is better to lower the burner air pressure. If the holding fltH is small, it means that the molten copper action on the shaft is weak, so in this case it is better to increase the burner air pressure. In the present invention, in addition to the holding amount H, the molten copper temperature Tk is also used as an input signal. For this reason, a thermocouple 20 is provided in the holding furnace 2 to measure the hot water temperature Tk. The temperature of the hot water may be measured during the exit route from the shaft furnace. This is the second input signal. As already mentioned, the first input signal is the tilting angle θ of the holding furnace, but this is replaced by the holding amount Hll by the tilting angle holding amount conversion as shown in the graph (c). Since the purpose of the present invention is to keep the holding amount H of the holding furnace 2 constant, the target holding ffiH. is determined in advance. In this example, the maximum value of the retained amount is 18t, which is the target retained amount H. is set to 10t, but the setting of HO is arbitrary. The water temperature is maintained at T1 at regular intervals (for example, every 3 minutes). iH
(Tilt angle θ) is sampled and the burner air pressure is determined based on this, so the suffix indicating the number of samplings is k, and this is added if necessary. Although k may be omitted, it means the same thing whether or not k is added. Target retention amount H. is subtracted from the current held amount H5, and the deviation EVk of the held amount is determined by difference calculation 11. EVk “Hu − Ho (1
). Next, use the difference calculation 12 to find the variation in the retention amount? E'/h = Evk − EVu−+
(2) Two H ■ - [1 k - 1 (3
). Deviation EV of retained amount and its temporal variation ΔE
Using V as an input signal, a fuzzy calculation is performed to find the amount of change ΔL in the combustion level. Is the hot water temperature new? It is used as a force signal, but not used in fuzzy calculations. The deviation of the retained amount is sometimes written as EV■ with the sampling time added, or it is sometimes abbreviated as EV. It can be simplified and also written as E. Similarly, the temporal changes are written as ΔEy, c1ΔEV, and ΔE, but they are all the same. In the fuzzy inference section, the retention amount deviation E1 and its time difference ΔE
For example, five membership functions are set from PB (positively large) to NB (negatively large). This is shown in Figure 2.
It is shown in Figure 3. The horizontal axis is E1ΔE, and the unit is 1 (}x). The vertical axis is 0-
It has a membership coefficient of 1. The membership function may be 3 or more. It may be trapezoidal or triangular, but here it is triangular and overlaps the adjacent membership function by about half. The central membership function ZO of the deviation E has a vertex at 0, −
It has a triangular shape with end points at 1.5 and 1.5. In other words, E≦-1.5teo, 1.5≦Eteote, -1
．． When 5≦E≦0, ZO=2E/3+ 1, O<E≦1.
5 (7) When ZO=-2E/3+1. Menhaship function PS (positive and small: positive
esmall ) is a function in which Z○ is shifted to the right by 1.5. NS (negatively small) is 1.5 to the left of ZO
It is a smooth function. The membership function PB (positively large) is semi-trapezoidal,
0 for E≦1.5, 0 for 1.5≦E≦3 (2E/3-1),
It has the form l where 3<E. NB (negatively large) is 1 at E minus 3, -3≦E<
-1.5 (-2E/3-1), -1.5≦E and O
This is the form. Such a shape can be changed arbitrarily. The above is the simplest example. The same applies to the time difference ΔE. However, strictly speaking, the unit of the horizontal axis is (1/sampling time). The combustion level change amount ΔL is the output of fuzzy inference,
This includes, for example, seven membership functions NB (negative and large),
It consists of NM (negative and medium), NS (negative and small), ZO1PS (positive and small), PM (positive and medium), and PB (positive and large). The number of membership functions is not limited to five, but may be three or more. A trapezoidal or triangular function may be used, but in this case, a triangular shape is used in which each half of the function is simply overlapped with the adjacent membership function. The membership function of the combustion level change amount ΔL is selected by a combination of the membership functions of the input retained amount deviation E1 time difference ΔE. This correspondence is made by a fuzzy rule formed by a combination of an antecedent (IF) and a consequent (THEN). Although there is some arbitrariness in how fuzzy rules are selected, FIG. 5 shows a control rule map as an example. Since the inputs are E1ΔE and each has five membership functions, there are 25 combinations of antecedents, and a maximum of 25 fuzzy rules can be created. A member bump function of E and a membership function of ΔE are arranged in the row and column directions. The membership function of ΔL is written at the intersection of these Membunoff functions. For example, when E is NB and ΔE is NB, ΔL is PB, which represents a fuzzy rule that says, "If E is negative and large, and ΔE is negative and large, then make ΔL positive and large." ing. In other words, "If the retained amount H is much less than H. and the retained amount H is decreasing significantly over time, immediately increase the combustion level." This judgment is the same as when operating a furnace manually. For example, if E is PB and ΔE is PB (bottom right corner), then Δ
L is NB. This is the opposite of the previous one.
fiH is H. If it is much larger and is increasing significantly over time, reduce the combustion level immediately. "That's what it means. Generally, even if E is large or ΔE is large, the retained amount is large or is becoming large, so it is better to lower the combustion level. Even if E is small or ΔE is small, it is better to raise the combustion level. Therefore, with respect to the rule map of FIG. 5, substantially the same membership functions with respect to ΔL are lined up in the diagonal direction downward to the left. And in the diagonal direction downward to the right, Δ
Membership functions that go negative with respect to L are lined up. In this way, there is a possibility of creating 5×5=25 fuzzy rules, but it is not necessary to actually adopt all of them as rules. The fuzzy operation will be explained with reference to FIG. The left column is the membership function of E, the second column is the membership function of ΔE, and the right column is the membership function of ΔL. For example, the retained amount deviation E is Ek in the graph of E in FIG. 6(a), and the difference ΔE is ΔE in the graph of ΔE.
Suppose that k. For Ek, the member sysop function of E that is not O is NB,
Assume that there are two types, NS and NS. The membership function of ΔE that is not 0 for ΔEk is zo
, ps. Let A1B be the point where the vertical line E=Ek intersects NB and NS. Let C1D be the point where the vertical line ΔE=ΔEk intersects with PS1Z○. Since the membership functions of NB1NS for E, Z○ and PS are related to ΔE, for ΔL, (NB, Z ○) → PB (NB, PS) → PM (NS, ZO) → PS (NS. PS) →Four membership functions of ZO must be considered. In FIG. 6(b), the membership function of NB is considered for E, the membership function of Z○ is considered for ΔE, and the membership function of PB is considered for ΔL. This corresponds to the fuzzy rule that says, "If the retained amount deviation is large and negative, and there is almost no change in the retained amount, make a positive and large change in the combustion level." The membership coefficient of NB to E1 is AJ and ΔE5
The member bump coefficient of ZO for is DL. I decided to cut PB at the smaller value of these, the combustion level Δ
The contribution of the member sop function PB of L is a trapezoidal portion of D1D2 or less. FIG. 6(C) shows NB for E, PS for ΔE,
It is about the membership function of PM for ΔL. Since PM is cut by the lower membership coefficient, P
The contribution of M becomes a trapezoidal portion below C and C2. This means, ``If the retained amount deviation is negative and large, and the time change is positive and small, then the combustion level should be changed to a positive medium degree.'' Figure 6 (d) shows that NS1 for E and ZO1 for ΔE.
This is the result of applying the PS membership function to ΔL. Since BJ is smaller than DL, the membership function of PS has a contribution of the trapezoidal part cut by +3, B2. This means that the retention amount deviation is negative and small, and the time change is O.
If so, change the combustion level by a small amount. ”. Figure 6(e) shows NS for E and PS1Δ for ΔE.
Does Z○'s member bump function correspond to L? It is something that has been set. This is a rule that says, "If the retained amount deviation is negative and small, and the retained amount change is positive and small, do not change the combustion level ΔL." Since BJ is smaller than CL, we get a trapezoidal contribution of B3B4 and below for ZO on combustion level. In this way, the values of the membership functions P81PM and PS1ZO of ΔL are obtained for the four fuzzy rules (the contributions of the remaining functions are all O), so their centers of gravity are determined and set as ΔLk. In this way, the combustion level difference ΔLk is obtained. Adding this to the previous combustion level Lk-1, the current combustion level Lk
get. Lk"Lk-1+ΔL■=ΣΔL (4).No matter what value E■ and ΔE■ take, membership functions belonging to these are combined in the same way, and first mln
It is possible to calculate ΔL■ by converting the membership function of ΔL into a trapezoid and finding the center of gravity of this trapezoid. The period of this various sampling and fuzzy calculation is, for example, 3 minutes. In the first inference, the relationship between the retained amount H1 and the combustion level L7 at the start of control is set in advance as a polygonal line function, and the initial value L1 of the combustion level is determined according to that relationship. In this way, the combustion level L at time k is determined. The steps up to this point are common to all burners A1B and A1C. From here, combustion level air pressure conversion is performed. This is given in advance by a line function as shown in graph (1). Since the combustion level L and the air pressure QAs= are almost proportional to each other, the curve slopes upward to the right. However, for burner A1B1C, L3 → QAh, QB
It is assumed that the relationships between i= and QCu are different. This is shown in FIG. Here, the horizontal axis is divided into seven combustion levels. The vertical axis represents the air pressure in each zone, and the unit is m m A Q. The same thing was recorded in the work table. The burner air pressure of each zone at each level is set as shown in Table 1, and by connecting each level with a straight line, the relationship between the combustion level and burner air pressure is given as a polygonal line function. In this way, the air pressure is determined by multiplying the uniformly determined combustion amount by a different coefficient for each burner. This is one of its characteristics. Another feature of the present invention is to correct the previously determined burner air pressure QA1QB1QC based on the hot water temperature Tk. For this purpose, temperature and coefficient conversion as shown in graph (e) is performed. This conversion also differs depending on the burner A1B1C. This coefficient K is calculated by multiplying the burner air pressure QA1QB, QC obtained in the previous process to obtain the true burner air pressure Q'A.
1Q'B N Q'C. This coefficient k also differs depending on A, B, and C. Q'Ax= QAk X KAk (5
)Q' Bk= QBkX KBI, (6)Q'
Ck=QCkXKCk (7) Figure 8 shows the values of the coefficients KA, , KBk1KCk depending on temperature. The water temperature is discretized at 5°C intervals, and coefficients are set for each zone. The solid line is the coefficient for the A zone, the broken line is the coefficient for the B zone, and the dashed line is the coefficient for the C zone. The coefficient is close to the value of the factor, but by multiplying it by this value, fine adjustments can be made. The same is shown in Table 2. The target value of the relationship '7% 1'J= between the second table water temperature and each zone coefficient is l125'c, and the coefficient is set to 1 in this vicinity. The sum of the three coefficients is set to be ], and the overall combustion level is determined by the above-mentioned Lk. This factor finely adjusts the downward combustion level distribution in the shaft furnace. In this example, when the hot water temperature Tk is low, the heating power of the lower burner A is increased to raise the hot water temperature. On the other hand, when the water wetness T3 is high, the firepower of the lower burner A is suppressed to lower the water temperature. ? However, this assumes an intuitive cause-and-effect relationship, and cannot always be said to be correct. Since the combustion level L is determined using the held ffiE and its temporal change ΔE, it may be better to reverse the trend of this coefficient. In any case, the coefficients KA, K due to N m temperature T y
B1KO settings can be made arbitrarily. By multiplying by 15, the air pressure in each zone is Q'Ak1Q'B
■, Q' 0.4 is determined. The subsequent steps are the same as in the conventional method, and these values are entered into the indicating controller 16. This passes through the positioner 17 and the changeover switch 5, becomes the valve operation signal 7, and drives the motor 6.
Adjust the opening degree of valve 8. Air supplied from the blower 9 is supplied to zones A, B, and C of the shaft furnace depending on the opening degree of the valve. In FIG. 1, the upper row shows a conventional device, and the middle row shows a device according to the present invention. Target retention amount H0 is 10t
When using conventional manual operation, the average value is 9.9G.
t, the standard deviation was 1.39t. According to the present invention, the average value is 9.97t and the standard deviation is 0.
．． It was 88 tons. In order to more intuitively show the stabilization of the retained amount, the actual temporal fluctuations of this are graphed in Figure 9 (conventional example).
It is shown in FIG. 10 (present invention). The horizontal axis is time and the vertical axis is the retained amount, but only the vicinity of 10t is shown. It can be seen that the temporal fluctuation of the retained amount is larger in the conventional example.

【Effect of the invention】

(1) Automatic control of fuel ti ffi makes it possible to save on furnace operation. (2) Quality can be stabilized by stabilizing the amount of melted copper and the temperature of molten copper. (3) Compared to manual combustion amount control, the fuel fffffi is operated frequently at short time intervals, resulting in an energy saving effect.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram of a shaft furnace automatic combustion apparatus according to the present invention and a conventional example. FIG. 2 is a graph showing an example of a member function of the retention amount deviation E. FIG. 3 is a graph showing an example of the membership function of the temporal change ΔE in the retained amount. FIG. 4 is a graph showing an example of the membership function of the combustion level change amount ΔL. FIG. 5 is a control rule map diagram of fuzzy rules that define the relationship between E1ΔE and ΔL. FIG. 6 is a diagram showing a procedure for determining the combustion level change amount ΔL from the value Ek1 ΔE of E1ΔE, by fuzzy calculation. FIG. 7 is a graph showing a predetermined relationship between combustion level L and each zone burner air pressure. Figure 8 shows hot water tM T and each zone coefficient KA, KB1KO
A graph showing a pre-given relationship. FIG. 9 is a graph showing an example of a temporal change in the amount held in conventional manual control (based only on the amount held 1). FIG. 1θ is a graph showing an example of a temporal change in the retained amount in control (fuzzy calculation using E1ΔE) according to the embodiment of the present invention. 16 3. 5. 7. 9. 11. 1 3. 1 5. 1 7. Shaft furnace, 2..., holding furnace indicator controller, 4... Posisilona changeover switch, 6... Motor valve operation signal, 8... Valve blower, 10... Valve opening difference calculation, 12. ．． ．． Differential fuzzy control, 14... Accumulation multiplication, 16... Indicating controller positioner, 20. ．． 、． Thermocouple Akira - Hiroshi Matsuyama
Masashi Shima Kitake Moto Yasu Toshisuga Yoichi Hara 2nd figure Membership function of deviation E 3rd figure Membership function of time change ΔE 4th figure Menobassive function of combustion level change ΔL 5th large negative to medium negative Small, almost zero, positive, small, positive, medium, positive, dog Figure 7 Relationship between combustion level and burner pressure Combustion level Figure 8 Relationship between the hot water pump and the coefficient of each zone 111'. Fumiaki Shida, Director-General of the Government Office 1. Indication of the case Patent application No. 1-301189 2 Title of invention Automatic combustion device for shaft furnace applying fuzzy theory 3. 71n Correct Seriyoj ``Seki with the matter 1 Series 4.1 Applicant's residence 4-5-33 Kitahama, Chuo-ku, Osaka Name (213) Sumitomo/U
Ki Kogyo Co., Ltd. Representative LI, Tetsuro Nagakawa 4th generation Director *537

Claims (1)

[Claims] A holding device consisting of a shaft furnace that has a plurality of burners A, B, etc. and melts copper, and a holding furnace that holds molten copper tapped from the shaft furnace. This is a device that automatically controls the combustion amount of the burner of a shaft furnace for the purpose of stabilizing the amount held in the furnace and the temperature of the molten copper. The output signal is the valve operation signal of the electric valve, and includes a preprocessor section, a fuzzy inference section, a postprocessor, and a hardware section. , Calculates the holding amount from the tilt angle according to the polygonal line function, calculates the deviation E between the preset target holding amount and the current holding amount, and calculates the change ΔE in the deviation E in the preset control cycle. , and the fuzzy inference part calculates from PB (positively large) to NB (negatively large) for E and ΔE.
Set a membership function of 3 or more to The amount of change in combustion level Δ for each combination of and ΔE
Set a rule map that gives During inference, the relationship between the retained amount and combustion level at the start of control is set in advance as a line function, the initial value of the combustion level is determined according to that relationship, and the combustion level is determined by multiplying this initial value by the inference result. The post-processor section sets the relationship between the combustion level and the burner air pressure of each zone in advance as a polygonal line function, and calculates the air pressure P_A of each zone from the combustion level determined by the fuzzy inference section according to the polygonal line function. , P_B..., and the coefficient that changes stepwise depending on the molten copper temperature (water temperature) is A,
Coefficients A_K, B_ that should be set in advance for each zone B and C and applied to each zone from the current water temperature
Calculate the value of K... and multiply the air pressure P_A, P_B... of each zone by the coefficients A_K, B_K... obtained in the previous process, and the air pressure of each zone obtained by this is the final The control result is output to the indicating controller, and the hardware section outputs a valve operation signal for the electric valve so that the burner air pressure obtained by the post processor section is obtained by the indicating controller and positioner. A shaft furnace automatic combustion device that applies the characteristic fuzzy theory.