JPH02156017A - Combustion control method for continuous type heating furnace - Google Patents

Abstract

PURPOSE:To execute combustion control with high accuracy so as to assure a material extraction temp. and to reduce fuel flow rate by setting the fuel flow rate by a target heating-up curve from the actual record values determined from in-furnace data and information and further optimizing the target heating-up curve. CONSTITUTION:The present actual record values are calculated and updated from the actual records of a plant and the data on in-furnace materials by a 1st actual record processing function 1 of the continuous type heating furnace having plural furnace zones. The fuel flow rate set values of the respective zones are determined and outputted in such a manner that the respective material temps. do not fall below the values on the target heating-up curve in a 2nd fuel flow rate setting, calculating and outputting function 2 in accordance with the actual record values. Further, the simulation of the heating furnace is executed while the fuel flow rate set values of the respective zones are determined by the method similar to the 2nd function 2 in a 3rd target heating-up curve optimizing function 3. The optimum target heating-up curve to minimize the evaluation function is determined while the values of parameters are corrected by using a nonlinear optimization method and is outputted to the above- mentioned 2nd function 2.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to combustion control of a continuous heating furnace having a plurality of furnace zones.

[Conventional technology]

In the combustion control of a continuous multi-zone heating furnace, the method of optimizing the temperature rise curve (temperature rise pattern) of the material online is as follows:
In recent years, several methods have been proposed as flexible and effective methods for optimizing operations, such as ensuring the firing quality for a variety of different material charges and saving energy.

A typical example is JP-A-61-1990.
In the method described in No. 16, the heating furnace is divided into n meshes in the furnace length direction, and for each material in the furnace, the fuel flow rate W when present in the mesh is determined.

The amount of change ΔWl from the current value and the average temperature of the material at the time of extraction 3. Uniformity (temperature difference between surface and inside) T's.

And, thirdly, the relationship between the mesh furnace temperature T and the amount of change from the current value of i is approximated by a linear equation to calculate t'r's.

Under the constraint on T, the fuel Wgo for each material is
, y (k=1.2.・'', n) are determined using linear programming. From this WxoPT, the furnace temperature T' vaci) for each zone for each material is determined (i is the material index).
The furnace temperature for each zone is determined as the weighted average value for each material.

On the other hand, in Japanese Patent Application Laid-Open No. 58-21.0i20, the furnace temperature is optimized instead of the fuel flow rate, but linear programming is used.

[Problem to be solved by the invention] Below, problems related to JP-A-61-199016 for optimizing fuel flow rate will be explained, but JP-A-58-210120
The problem definition is the same even when optimizing the furnace temperature.

Now, in the conventional continuous heating furnace control method as described above,
The relationship between the amount of change in the fuel flow rate from the current value and the amount of change in each of the material average temperature, soaking degree, and furnace temperature during extraction is approximated by a linear linear equation, but these relationships are essentially Since it is nonlinear, the error increases as the amount of change increases. Therefore, when the size of the material, the scheduled furnace time, and the extraction target temperature vary, the amount of change described above also increases, leading to an increase in the error of linear approximation. Since it is necessary to increase the fuel flow rate to a certain extent to finely change the fuel flow rate, the parameters to be optimized for all materials in the reactor (the number of fuel flow rate changes for each material side category will increase, and if linear approximation is not performed, it will remain non-linear. When attempting to perform optimization using , the amount of calculation becomes enormous, making it difficult to perform online.

On the other hand, in the operation of a lump-heating furnace, the operating schedule (remaining in-furnace time of the material, etc.) is often changed depending on the work status of the previous and subsequent processes. ), schedule changes have a large influence on the optimum parameter values, and optimization calculations must be performed every time the schedule is changed, resulting in a large calculation load. Furthermore, since the optimization calculation requires a certain amount of time, the action of changing the operation amount in response to the schedule change is delayed by this amount of time, reducing the effectiveness of the optimization.

[Means to solve the problem]

The present invention solves the problems of conventional control as described above, and sets a target temperature rise curve (target temperature rise curve) for each material from charging to extraction on the material side. The fuel flow rate is set at a predetermined value so that the temperature of each material in the future does not fall below the target temperature rise curve for a predetermined time within the specified period of time, or the temperature of each material becomes the target temperature rise curve on average or preferentially for a specific material. At the same time, each material is divided into several groups with similar materials that are close together as one group, and common parameters of the target temperature rise curve that take a common value within each group (for example, each zone (e.g. material temperature increase rate) is determined by a nonlinear optimization method such as the steepest descent method so as to minimize a predetermined evaluation function value.

The value of the evaluation function is obtained by simulating the heating furnace operation until a predetermined material is extracted using a process model (heating furnace simulator) that includes a nonlinear heat balance equation and a heat conduction equation and can be calculated at high speed. Since no linear approximation of the equation is performed here, unlike the prior art, no errors occur due to linearization. In addition, the function of the common parameter of the target temperature increase curve, which is the parameter to be optimized, is significantly smaller than the number of future fuel flow rate time series values in the conventional technology, making the quantity optimization calculation more efficient.

On the other hand, when changing the operating schedule of a heating furnace, if the target temperature rise curve is expressed as a function of the remaining furnace time, for example, if the remaining furnace time of a certain material is changed, the temperature of that material after a predetermined time will be Since the target value is also automatically changed on the target temperature increase curve, there is no need to change the common parameter values that define the target temperature increase curve much. The same is true when the material extraction target temperature is changed.In the end, in the present invention, changes in the operating schedule have little effect on the optimal values of the parameters to be optimized, and therefore optimization calculations are performed every time the schedule is changed. There is no need to do this, reducing the calculation load, and
Changing the amount of operation when changing the schedule can be done immediately without waiting for optimization calculation, and the effect of optimization is maintained.

[Effect]

The present invention will be explained in detail below.

The target plant of the present invention is a multi-zone continuous heating furnace,
An example is shown in FIG. In the heating furnace, materials with different material data (thickness, width, length, material, charging temperature, extraction target temperature, in-furnace time, predicted extraction pitch value with respect to preceding materials, etc.) are transferred from the charging side to the extraction side. While conveying to the extraction target temperature, the temperature is increased to the extraction target temperature.

The present invention is a method for controlling such a heating furnace, and its functional block diagram is shown in FIG.

Among these, the first actual performance processing function (first function) is started at a predetermined period, and the plant actual performance φ [furnace temperature of each zone of the heating furnace (
By inputting the observed values of gas temperature), fuel flow rate, and material data including the material position inside the furnace and of the materials scheduled to be charged into the furnace, you can calculate the temperature in the thickness direction of each zone furnace wall and each material that cannot be directly observed. Calculate and update the distribution actual value (current value), and calculate the plant actual value φ
In addition to φ', in order to deal with changes in the characteristics of the heating furnace, the estimated value θ of the unknown parameter in the heating furnace simulator, which is an important process model in the present invention, is set based on the actual plant performance φ'. At the same time, the second fuel flow rate setting value calculation function (second function) is activated. Here, the sampling period is made equal to the simulation period Δt of the heating furnace simulator. φ updated in the first function
', θ are used in other functions.

Next, the second function is started from the first function, and the fuel flow rate setting value F! of each combustion zone (zone where fuel is burned) of the heating furnace is activated! v is calculated from the current values of each zone furnace temperature and material temperature, and the target temperature rise curve of each material, and is output to the plant.

Finally, the target temperature rise curve optimization function (third function) divides the materials into several groups in the order of extraction, and calculates the common parameters (the same value is taken for the same group) of the target temperature rise curve of each material. The value is determined such that the value of a predetermined evaluation function is minimized. The obtained value A* is used in the second function.

Here, the evaluation function is set according to operational objectives such as ensuring the quality of the baked clay material and saving energy.

The value of the evaluation function is determined by simulating the operation of the heating furnace from the current moment until a predetermined material is extracted using a heating furnace simulator that is a process model. This third function does not need to be activated in synchronization with the first and second functions, and may be activated independently at a predetermined cycle.

Below, a heating furnace simulator, which is a basic process model used in the present invention, a furnace temperature prediction model, which is used to regulate the upper and lower limits of the furnace temperature, a material temperature increase model, which is used to calculate the fuel flow rate set value, and a material target temperature increase curve. , and the evaluation function used for target temperature rise curve optimization, and then the first to
The third function will be explained in detail.

(1) Heating Furnace Simulator The heating furnace simulator simulates the operation of the heating furnace at a predetermined time interval Δ.
From the initial values (values at the initial time) regarding wall temperature, fuel flow rate, and material data, and the values of each zone fuel flow rate for each ΔL after the initial time, the furnace temperature, wall temperature, and wall temperature for each ΔL after the initial time are calculated. Calculate material temperature.

If this initial time is set as the current time, future operations can be simulated. Although the heating furnace simulator is a well-known technology, it plays an important role in the present invention and is also related to the parameter sequential estimation described later, so the processing steps for time determination will be explained in 1) to 3). Also, in the following description,
Let Δt be the unit of time. (i.e. Δt=1) ) Simulate the material transport that occurred during time (t-Lt). If there is any material extracted during this period, it will be removed from the in-furnace material data. If there is a gap at least a predetermined interval on the charging side, the material to be charged (or a dummy material of constant size and material) is charged.

2) The temperature of each zone at time (t-1) is determined from the fuel flow rate of the zone and the zone on the extraction side, the wall temperature of the zone, the material surface temperature, etc. at time (t-1), and the heat balance within the zone is calculated as shown below. Find it using 2.

QA50QM1 +QGi-1+QGmu Qsi Q
wt QLi = 0...(1) Q□=Σ(σ・φ6.・S,・(TF positive j' Tsi
j'))・・・・・・(2) Qwi= σ ・ φ ew ・ Swi ・
(Tyi'T-t') - (3) Q Li
=λ61 ”rri+λ■ ・・・・・・
(4) where l: band index (i=1.2...; the band at the extraction end is t=1); j: index of the material in the first band; Qat: sensible heat of preheated air; Q , 1L: Calorific value of input fuel; Qei-+: Sensible heat of inflow waste gas (other than soaking zone); Qai:
sensible heat of the outflowing waste gas; Qsi: heat input to the material; Ql: heat input to the furnace wall; QLi: heat loss and correction term; σ, Stefan Boltzmann constant; φ. ,・Overall heat transfer coefficient (gas-material); φ. : Overall heat transfer coefficient (gas-furnace wall); S, : Material surface area; So : Furnace wall area: T□ : i-th zone furnace temperature (gas temperature); Tsi, : surface temperature of j-th material; TWi : furnace Wall temperature; λAi+λ■: heat loss coefficient, the value is estimated one by one using actual operation data (described later).

When the number of zones included in the heating furnace is n2, equation (1) is established in each zone, so there are a total of n2 heat balance equations, and on the other hand, there are also nz unknown variables at each zone furnace temperature. Therefore, the furnace temperature TFi can be obtained by combining the heat balance equations, but since this equation is a quartic equation regarding TFi, the solution is obtained by the well-known Newton-Loughran method. Figure 3 illustrates the concept of heat balance within the zone.

3) The thickness direction distribution of each zone wall temperature at time and the thickness direction distribution of each zone material temperature are calculated from the thickness direction temperature distribution at time (t-1) and the furnace temperature at time t as follows. It is determined using a well-known heat conduction equation that is discretized in the time and thickness directions.

is given by here T: Temperature (wall temperature or material temperature); t: time; Thickness direction coordinate: α: temperature conductivity; In: Boundary conditions.

Note that calculating the temperature distribution in the thickness direction using the above equation for all materials in the furnace requires a considerable amount of calculation, so a known weighted residual method is used to improve calculation efficiency.

A simulation is performed by repeating the above processes 1) to 3) while updating time t by 1.

(2)) Furnace temperature prediction model In the present invention, the manipulated variable for control is the fuel flow rate, but in actual equipment, the upper and lower limits of each zone furnace temperature are determined from the equipment or the quality of the materials. Therefore, the fuel flow rate set value must be set to a value that does not exceed these upper and lower limits for the furnace temperature, and for this reason, a model that expresses the dynamic relationship between fuel flow rate changes and furnace temperature changes is required. , newly developed as a furnace temperature prediction model. The model formula is shown below.

・・・・・・(6) However, ΔT, i (t) = T□(t) Tri(t
1) ΔF 1(t)=F 1(t)−Fi(t−
1) Here, i: Band index; T, ,: Furnace temperature; Fi: Fuel flow rate; art+cyi+, +(k = Q, i,...',
my) : Appropriate constant (value determined for each band based on actual data); m, : Appropriate positive integer (m, = 2 when Δt=2min
).

(3) Material Temperature Rise Model In the present invention, it is necessary to find a fuel flow rate for each material so that the temperature of each material matches the target temperature rise curve for that material after a predetermined period of time. For this purpose, a simple model that expresses the dynamic relationship between fuel flow rate changes and material temperature changes is needed, and a new material temperature increase model was developed. The model formula is shown below.

・・・・・・(7) However, ΔTsdt)=Tsj(t) Tsj(t 1)Δ
F, (t) = F, (t) −F, , (t −1)i:
Band index; j: material index (includes band No. 1); n: band index; T, ,: material temperature (average value in the thickness direction); 5Hj - material specific heat; 1,, = material thickness; bnj: nth Expressing the degree of influence of zone furnace temperature T5 on material j,
Constant determined by material position.

An example of bfij (n=1-1°i, i+1) in the first zone is shown in FIG.

aSi+ CSil+ 11' Constant (value determined for each band from actual data) m,: Positive integer (when Δt-2m1n, about m,!2 may be sufficient).

In addition, in the second term on the right side of equation (6), when i-1 (band at the extraction end) or i = n, (band at the charging end), there may be cases where n-0 or n = n2 + l. , in these cases, b *J "" CMk+ +%-〇.

(4) Target temperature rise curve for materials A target temperature rise curve is set for each material for the material that has been charged into the furnace and a predetermined number of materials that are scheduled to be charged.

This curve represents the target value for increasing the material temperature (average temperature within the material) during the period from material charging to extraction and is a function of time. When charging materials, the value of the target temperature rise curve should be set to the actual value (predicted value for the material scheduled to be charged) of the material charging temperature (average temperature inside the material at the time of charging), or Set the target value to a value close to the charging temperature (if it is not necessary to immediately raise the temperature from the time of charging, it is effective for energy conservation to set the target value slightly lower than the charging temperature).

On the other hand, if the remaining furnace time until material extraction is less than the predetermined ripening time, the target temperature rise curve is Equal to the extraction target temperature of the material. This aging time is determined by taking into account the charging temperature of the material, rolling specifications, etc.

From charging to the start of aging, a target temperature increase curve is completed by continuously connecting a plurality of straight lines or curves. Various parameters that define the target temperature rise curve (slope of straight line or curve,
Some of the values (such as aging time) are set as common values within the material group described later, and in the target temperature rise curve optimization process, a value (optimal value) that minimizes a predetermined evaluation function is determined.

An example of the target temperature increase curve is shown in FIG. In this example, the heating furnace has a preheating zone (PHZ), #1 heating zone (#IH) from the charging side.
2), #2 heating zone (#2Hz).

It consists of four zones: a soaking zone (SZ), and the zones other than PHZ are combustion zones (zones where fuel is burned). The target temperature rise curve is PH
z, #IHz is a quadratic curve, #2H2 is a cubic curve, 3
The distance from the two people to the start of ripening is represented by a straight line, and the tangential line between the #2Hz inlet and the outlet is smoothly connected (with continuous differential coefficients).

(5) Evaluation function In the operation of a heating furnace, it is necessary to comprehensively optimize several objectives such as ensuring the baking quality of materials and saving energy, and the evaluation function JIF for this purpose is set using the following formula.

Here, i: band index; j: material index (targets materials existing between the extraction end and a predetermined position on the charging side and is referred to as evaluation target material); band); sum related to the time until singing; ΔTstj: extraction material temperature deviation (difference between the average temperature of the j-th material at the time of extraction and the extraction target temperature (average - target)); T'stj: material meat uneven heat (extraction temperature deviation in the j-th material (surface-center); r, , r, appropriate function; Fi: fuel flow rate in the first zone; W+, Wg, W3: extraction temperature deviation, material meat uneven heat, and Common weighting coefficient for fuel flow rate; WlBj, Wz3j: Material-specific molding coefficient for the j-th material The first to third terms of the above evaluation function evaluate the extraction temperature deviation of the material, the uneven heat of the material meat, and the fuel flow rate, respectively. It represents.

The common weighting coefficient is set to an appropriate value according to the operational purpose (Is the emphasis on material baked goods or energy conservation?), and the molding coefficient for each material is determined, for example, if the Determine the value for the material, taking into consideration whether it is necessary or not.

Function f,,f! are respectively 1ΔT$Ej l r T'
In order to prevent the 5tJO value from increasing, it is determined that the function value increases as these values increase.

This evaluation function is used for target temperature rise curve optimization (third function), which will be described later.

(6) Actual performance processing function (first function) Simulation period Δt (~1
) and performs the following processes 1) to 3).

l) Updating the temperature distribution in the thickness direction of each zone wall temperature and each material temperature,
From the previous value of the temperature distribution and the actual value of each zone furnace temperature, the temperature distribution is updated and the current value is determined using equation (5) of the heating furnace simulator.

2) Sequential estimation of parameters Heat loss coefficient λ, & in furnace simulator.

λ5i (i is a band index) is sequentially estimated using performance data every time performance data is obtained at a predetermined period. The equation for updating the estimated value at the time of day is shown below.

ai(t)=ai(t 1) +k(t)(Jffi
(t) al(t'1)'xi(t))...
・(9) Here, QLi(t)=QAt(t)+Qxi(t)+
Qci-+(t) + QGi(t)Qsi(t)
Qwt(t) --00)at(t)
= [λAi+ λ, il” (parameter estimated value vector)
λl i (t) is the estimated value at time t of λ■, λ, and 5, respectively, Fi is the fuel flow rate in the first zone, and QLi(t)
is the actual loss heat value at time t, and is obtained by calculating the value on the right side of equation (00), which is obtained by converting equation (1), using the actual data at time t.

On the other hand, k (t) is a parameter correction gain vector (two-dimensional) for the estimation error of QLi (the value in parentheses in equation (9)), and is calculated using a known method such as the recursive least squares method.
Find the value at the time. With this parameter sequential estimation function, a heat loss coefficient that reflects actual data is always determined, and the accuracy of the heating furnace simulator is maintained even if the characteristics of the furnace change.

After this process is completed, the second function is activated.

3) Start function 2.

(7) Fuel flow rate set value calculation/output function (second function) A schematic flowchart of this function is shown in FIG. 6, and the processing procedures according to the diagram are described in 1) to 5) below.

1) Initialize simulation time t to current time t.

2) In each fuel zone in the furnace, calculate the fuel flow rate setting value Fi(t) for that zone in order from the extraction side using the following steps a) to d).

a) The band (first band) and the band on the charging side (the first band)
The material temperature (average temperature) after a predetermined time N3 for the material existing in a predetermined range on the extraction side of (i+1) band) Tsj
(t + Ns) is the target temperature rise curve value T sj of the material
” (t + N 3)
First zone fuel flow rate set value F st,j(
j) (j is the material index). Here, the (i
+1) The reason why the zone extraction side slab is also included is because the i-th zone fuel flow rate also affects the temperature rise of the material at this position.

The value of N is set to a small value in the soaking zone because it is highly necessary to match the material temperature to the target as quickly as possible. In order to avoid large fluctuations in the fuel flow rate, which is the manipulated variable, when the target value fluctuates due to fluctuations, etc., it should be made larger than the soaking zone. Further, the target trajectory of the material temperature change between Tij(t) and T sj'' (t + N3) is expressed by a straight line connecting the two as shown in FIG. 7, thereby suppressing large changes in the fuel flow rate.

Various methods can be considered for calculating Fsi,j(t), but the general method is effective in ensuring the ability to follow the slab temperature to the target trajectory, ensuring control stability, and preventing excessive fluctuations in fuel flow rate. Predictive control
d Predictive Control). Using this method, the following function JFL
An operation that minimizes 1tu(t)(=Fst.

j (t) ) can be obtained using the material temperature increase model (Equation (7)).

JFL (1'L, Nz, No) ・・・・・・(11) However, V (t + k) = Tsj (t + k) (k
=N++N+ + 1. ..., N2) Δu (10k
) = u (to 10k) −u (to 10 to 1) (k=
o, 1. ..., NLl-,) represents fluctuation. Therefore, control is performed to reduce both of them.

Note that the calculation becomes easier if the control is stable. Handling amount t's
The specific calculation procedure for determining t,j(t) is well known;
Omitted.

b) The first zone fuel flow rate setting value Fi(t) is set for a predetermined time (N
3) It is determined so that the future temperature of each material does not fall below its target temperature rise curve, or it is determined so that each material temperature on average or preferentially for a specific material becomes the value of the target temperature rise curve.

The calculation formula for F, (t) in the former case is expressed by the following formula, where F, (t-1): first zone fuel flow rate set value at time (t-1) λ: manipulated variable fluctuation weighting coefficient (λ≧0) NI+NZ+NU' Positive integer (N+≧N,) yr: The first term of the target trajectory evaluation function JFL of material temperature (y) represents the deviation of the material temperature from the target trajectory y1, and the second term The fuel flow rate WLi,j represents the weight of the j-th material on the first zone, and the value is determined as shown in FIG. 8 depending on the position of the j-th material in the furnace. In the figure, dk+1(k-1,...,
5) is an appropriate constant determined for each band.

C) Modify the fuel flow rate set value Fl(t) for the first zone so that the i-th zone furnace temperature does not exceed predetermined upper and lower limits determined by equipment or material quality.

To do this, first use the furnace temperature prediction model (6) to predict the future furnace temperature at 4 points in time from the furnace temperature, actual fuel flow rate, and Fsi+j (1) up until now. If the temperature exceeds the upper or lower limit, the fuel flow rate set value is corrected again using the prediction model so that it matches the upper or lower limit of the furnace temperature.

d) If the fuel flow rate setting value corrected in C) above exceeds the predetermined upper and lower limits on the equipment, make corrections so that it does not exceed it.

3) If the simulation time is also the predetermined time (t+N4) for the completion of the simulation, proceed to 5).

In other cases, advance t by one and proceed to 4).

4) The furnace temperature at time t of each zone from the extraction side to the charging side, the thickness direction temperature distribution of each zone wall temperature and each material temperature, and these previous values at time (t-1) and 2) From the fuel flow rate of each zone obtained in step 1, use a heating furnace simulator to find it, and return to step 2).

5) Fuel flow rate setting value F that is actually output in each combustion zone
xvt (tJ, the current fuel flow rate actual value and the set value F5 (mujk) (k = 0.1...N4) obtained from 2), to suppress large fluctuations in the set value. It is determined by appropriate filtering (smoothing) processing and output to the fuel flow controller.

(8) Target temperature rise curve optimization function (third function) A schematic flowchart of this function is shown in Figure 9, and the processing steps according to the diagram are shown in 1) ~
2).

1) Optimization pre-processing a) Material group division: At a predetermined period, the materials in the furnace and the materials scheduled to be charged into the furnace are searched in order from the extraction side in the furnace length direction, and similar groups are searched according to the division criteria below. Divide the whole into several material groups, with the materials in the same group as much as possible.

Group division criteria: i) to iv below between adjacent materials
) is established, group division is performed using this as the boundary of material groups.

i) Each band time is thick (different).

i:) Charging temperature is significantly different.

Ij) Extraction target temperature is thick (different) iv) The length of the group exceeds the predetermined length.

b) Initial setting of common parameters The values of the common parameters (see the explanation of the target temperature increase curve above) that determine the target temperature increase curve for each material are initialized for each material group. At the time of control start-up, predetermined initial values are set equally for all groups, but in other cases, they are determined by the following averaging process.

Σa'xij Matadashi i is the group index, j is the material index in the first group, axi is the common parameter a,
Initial setting value (value to be sought) in the i-th group + "xi"
j represents the current value of the common parameter for the j-th slab of the i-th group (i.e., the value obtained in the previous optimization process), and Σ represents the sum of the materials in the first group (
If the material belonging to the first group this time was divided into two or more groups last time, the value of a'xLJ differs depending on j).

2) Search for the optimal value of the common parameter Here, we use a known method for nonlinear optimization to find the point where the value of the evaluation function in equation (8) is the minimum (i.e., the optimal value of the common parameter) in the common parameter space. Search using a steepest descent method.

In order to obtain the value of the evaluation function, the heating furnace simulator is Simulate the operation using This simulation is the same as the simulation for calculating the flow rate setting value shown in FIG.

The search consists of the following directional search and straight line search.

a) Direction search In the common parameter space, find the direction in which the gradient of decreasing value of the evaluation function is maximum (direction of steepest descent). To do this, it is necessary to find the gradient of the change in the evaluation function value in each direction in space, which means moving the common parameter value by a predetermined minute amount in that direction and finding the evaluation function value at that point.
This is done by calculating the amount of change in value from the original position.

As a result of the direction search, if the direction of steepest descent is found, perform the following b); if not, consider the current point to be the optimal value that minimizes the evaluation function value (that is, optimal value A). Then, the process ends.

b) Straight line search Once the direction of steepest descent is found as a result of the direction search, find the point where the evaluation function value is minimum in that direction. For this search, a known method such as the Arm1jo method is used, and detailed explanation will be omitted. After the minimum point of the evaluation function is found, the direction search and straight line search are repeated again until the optimum point is found.

〔Example〕

Hereinafter, as an example of the present invention, simulation results using actual operation data will be described.

The target plant is a continuous heating furnace for billet slabs in a continuous hot rolling mill of a steel mill. The heating furnace has a furnace bottom zone, a first heating zone (#IH,z), and a second heating zone (#IH,z) in order from the charging side.
#2Hz), third heating zone (#3Hz).

It is divided into five zones: a soaking zone (Sz), and the extraction side from tt2Hz is a combustion zone. Figure 10.11 shows the results of actually operating this heating furnace in the conventional manner and the results of simulating the operation using the present invention for the same material as in this actual operation.

First, Figure 10 shows the material temperature at the time of extraction (material average temperature)
A comparison of the time course of is shown together with the extraction target temperature. The materials targeted here have a thickness of about 250 pieces, a width of 760 to 1460 am, and a charging temperature of 50 to 9.
It is ordinary steel at 00℃. As shown in the figure, the extraction target temperature varies greatly between 1100°C and 1230°C. It is desirable that the extraction temperature of each material does not fall below this target temperature.

However, as can be seen from the figure, in the actual operation results using the conventional method, the material after the target extraction temperature rose,
Insufficient baking is observed where the extraction temperature is lower than the target temperature. On the other hand, according to the simulation results according to the present invention, undercooking does not occur, and when the extraction target temperature decreases, the extraction temperature decreases more quickly than in the conventional method, thereby preventing the material from being overcooked.

Next, FIG. 11 shows a comparison of the transition of material uneven heat during extraction. It is preferable for the polarized heat to have a large positive value (in the conventional method, several peaks of the polarized heat appear, but in the present invention, the values of these peaks are all small.

On the other hand, regarding the fuel flow rate in this example, the flow rate was reduced by about 10% in the present invention compared to the conventional method. In the present invention, the flow rate by zone decreases greatly in the second heating zone on the charging side, decreases by a small amount in the soaking zone, and increases in the third heating zone.
Ta. This serves the purpose of reducing waste gas losses and reducing material heat imbalance during extraction.

As described above, in this example, the present invention showed more favorable results than the conventional operation in all of the extraction temperature of the material, the uneven heat of the material during extraction, and the fuel flow rate.

〔Effect of the invention〕

As described above, according to the present invention, by simulating the future operation of a heating furnace using a heating furnace simulator that includes nonlinear equations, each material is The target temperature rise curve of the furnace is optimized online, and the values of unknown parameters in the heating furnace simulator are sequentially estimated online using actual data, making it possible to optimize operations with high accuracy. It is possible to achieve effects such as ensuring material extraction temperature, suppressing material-related uneven heat during extraction, and reducing fuel flow rate.

[Brief explanation of the drawing]

Figure 1 is a functional block diagram of the present invention, Figure 2 is a diagram showing an example of a target plant, Figure 3 is a diagram showing the heat balance within the zone, and Figure 4 is a diagram showing the degree of influence of furnace temperature on materials. Figure 5 is a diagram representing the target temperature rise curve of the material.
Figure 6 is a flowchart of fuel flow rate setting value calculation, Figure 7 is a diagram representing the target trajectory of material temperature, Figure 8 is a diagram representing weighting coefficients depending on material position, and Figure 9 is a target temperature rise curve optimization process. FIG. 10 is a flowchart showing the transition of the material extraction temperature in the example, and FIG. 11 is a transition diagram of the material uneven heat during extraction in the example. Applicant Nippon Steel Corporation Patent Attorney Minoru Aoyagi Wt1 Figure 2 Figure 3 N4 Illustration Figure 7

Claims (1)

[Claims] 1. In combustion control of a continuous heating furnace having a plurality of furnace zones, plant results (observed values of each zone furnace temperature and fuel flow rate) and data of each material in the furnace are collected at a predetermined period. (material position, thickness, width, length, material, charging temperature actual or predicted value, extraction target temperature, in-furnace time or predicted extraction pitch between the preceding material, etc.) and directly observe the information. Each zone wall temperature cannot be
A performance processing function (first function) that calculates and updates current performance values such as the temperature of each material; to ensure that the material temperature does not fall below the value (target value) on the target temperature rise curve (target curve of material temperature transition from charging to extraction), or that the material temperature is average or preferential for a specific material within the same zone. The fuel flow rate setting value calculation/output function (second function) calculates and outputs the fuel flow rate setting value for each zone so that the target value is achieved, and the optimization of the target temperature rise curve for each material. The second function is activated at a predetermined period, either following the second function or independently, to control the operation of the heating furnace in the furnace or until the predetermined material to be charged is extracted in the future in the same manner as the second function. By simulating the fuel flow rate set value for each zone at every moment of By repeating the calculation of the evaluation function value through the above simulation while modifying the values of the parameters that define the target temperature rise curve using a nonlinear optimization method such as the steepest descent method, the value of the evaluation function can be minimized. A combustion control method for a continuous heating furnace, comprising: a target temperature rise curve optimization function (third function) for determining an optimal target temperature rise curve for each material. 2. In calculating the fuel flow rate setting value for each zone in the second and third functions described in claim 1, the temperature of each material in the furnace is set in the future for a predetermined period of time with the value on the target temperature rise curve (target value). The value of the fuel flow rate for each material required for A combustion control method for a continuous heating furnace, characterized in that a maximum value or a weighted average value is set as a fuel flow rate setting value for that zone. 3. In calculating each zone fuel flow rate setting value in the second and third functions described in claim 1, changes in each zone furnace temperature due to changes in each zone fuel flow rate setting value from now onwards in the future are calculated by combining the fuel flow rate change and the furnace temperature. A furnace temperature prediction model that expresses dynamic changes with temperature changes is used to predict the furnace temperature. A combustion control method for a continuous heating furnace, characterized in that the fuel flow rate setting value is corrected using the furnace temperature prediction model so as to be equal to the lower limit value. 4. In the second and third functions according to claim 1, the target temperature rise curve for each material is set as a function of the remaining furnace time, and when charging the material, the value is set at or near the charging temperature. , and while the remaining furnace time after charging is greater than a predetermined value, the target temperature increase curve is a continuous connection of multiple straight lines or curves determined by predetermined parameters such as the temperature increase rate of each zone. When the remaining furnace time is less than the predetermined ripening time, the target temperature rise curve is made equal to the material extraction target temperature in order to reduce the temperature deviation within the material, and the target temperature rise curve is modified using the above parameters. A combustion control method for a continuous heating furnace, characterized in that the combustion control method is performed by correcting the value of. 5. In the third function according to claim 1, in the heating furnace, several consecutive materials having similar values such as each band time, charging temperature, extraction target temperature, etc. are grouped together as the same group, The target temperature increase curve parameters according to claim 4 are characterized in that by using common values for materials in the same group, the number of parameters is reduced and the search for the optimal target temperature increase curve is made more efficient. , Combustion control method for continuous heating furnace. 6. In the third function according to claim 1, the value of an unknown parameter (heat loss coefficient) in a heating furnace simulator, which is a process model used for simulating heating furnace operation, is calculated from values such as furnace temperature, fuel flow rate, material temperature, etc. A combustion control method for a continuous heating furnace, characterized by sequentially estimating from actual values using a method such as a sequential least squares method.