JPH0914627A - Mixed fuel burning ratio estimating and combustion control method for fluidized bed incinerating furnace - Google Patents

Abstract

PURPOSE: To contrive the restriction of generation of CO and NOx by a method wherein the mixed fuel burning ratio of matters to be incinerated, which are different in the properties thereof, is estimated at real time while the set values of furnace outlet gas temperature, fluidizing air flow rate and secondary air flow rate are lead out employing the estimated value of the mixed fuel burning ratio to effect effective combustion control in accordance with the supplying amount of matters to be incinerated. CONSTITUTION: Upon burning matters to be incinerated in a fluidized bed incinerator, the mixed fuel burning ratio of the matters to be incinerated, which are different in the properties thereof, is estimated at real time employing a measuring signal in the fluidized bed incinerator and signal processing by a neural network 28 whereby a furnace outlet gas temperature, a fluidizing air flow rate and a secondary air flow rate are lead out in accordance with the supplying amount of matters to be incinerated to determine an operating amount from these values and control the combustion. The fluidizing air flow rate GA1 , a fluidizing air temperature TA1 , the secondary air flow rate GA2 , a secondary air temperature TA2 , a water pouring amount GSP in layer, the furnace outlet gas temperature TC, a layer temperature TB and a furnace outlet oxygen concentration O2 are employed as the measuring signals.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for estimating a co-firing rate of an incinerated material in real time using a neural network in an industrial waste incinerator provided with a fluidized bed and controlling combustion.

[0002]

2. Description of the Related Art In a fluidized bed industrial waste incinerator, the co-firing rate varies depending on the supply amount of the incineration material, so that it is necessary to control the combustion in accordance with the co-firing rate. However, heretofore, a constant co-firing rate derived from the design value of the incinerator has been used.

[0003]

When a constant value is set as the co-combustion rate as in the prior art, even if the actual co-combustion rate changes with a change in the supply amount of the incineration material, the flow rate is not reduced. Since the set values of the air flow rate, the secondary air flow rate, and the furnace outlet gas temperature do not change, the control performance is limited. The present invention has been made in view of the above points, and an object of the present invention is to improve a control performance by estimating a co-firing rate in real time by using a measurement signal such as a temperature and an air flow rate and a signal processing by a neural network. It is to provide a method.

[0004]

Means and Actions for Solving the Problems In order to achieve the above-mentioned object, the method for estimating the combustion rate and controlling combustion in a fluidized bed incinerator of the present invention, when burning an incinerator in a fluidized bed incinerator, By using the measurement signals of the fluidized bed incinerator and the signal processing by the neural network to estimate the co-firing rate of the incinerated substances with different properties in real time, the furnace outlet gas temperature according to the supply amount of the incinerated substance, The air flow rate and the secondary air flow rate are derived, the manipulated variable is determined from these values, and combustion control is performed. As the measurement signal, a flow air flow, a flow air temperature, a secondary air flow, a secondary air temperature, a water injection amount in a bed, a furnace outlet gas temperature, a bed temperature, and a furnace outlet oxygen concentration are used.

A neural network is a neural network model (Neural Network Model) that mimics the neural cells and their connection mode in the brain and the like.
1), which can be used for solving an optimization problem. Using measurement signals such as temperature and air flow rate and signal processing by a neural network, the co-firing rate of incinerated materials with different properties is estimated in real time. The set values of the furnace outlet gas temperature, the flow air flow rate, and the secondary air flow rate are derived, and the combustion control according to the supply amount of the incineration material is effectively performed.

[0006]

EXAMPLES Hereinafter, the present invention will be described in more detail with reference to examples, but the present invention is not limited to the following examples, and can be implemented with appropriate modifications. FIG. 1 shows a schematic configuration of a fluidized bed industrial waste incinerator (hereinafter, referred to as a fluidized bed furnace) 10. This fluidized bed furnace 1
At 0, the primary air (fluidizing air) 12 is supplied to the wind box 14 by the primary air blower to cause the fluidized medium of the fluidized bed 16 to flow, and the incinerated materials having different properties are introduced into the fluidized bed 16. The incineration material is gasified and burned, and the unburned gas generated by this is supplied to the freeboard section 18 by the secondary air blower to supply the secondary air 20 to complete combustion. Further, cooling water 22 is supplied to the fluidized bed furnace 10 in order to control the temperature of the fluidized bed 16 within a certain range. 24 is an air distribution plate.

In the fluidized-bed furnace 10 shown in FIG. 1, the equations for the energy balance of the fluidized bed portion, the energy balance of the freeboard portion, and the air ratio are represented by equations (1) to (3). _{C B W B dT B / dt} = H U1 K 1 G R1 + H U2 K 2 G R2 + H U3 K 3 G R3 + C PA T A1 G A1 + Q BI -Q BO -C PG T B {G A1 + (22. _{4/18) G SP} -600G SP (} 1) C PG T G {G A1 + G A2 + (22.4 / 18) G SP} = H U1 (1-K 1) G R1 + H U2 (1-K _{2) G R2 + H U3 (} 1-K 3) G R3 + C PG T B {G A1 + (22.4 / 18) G SP} + C PA T A2 G A2 (2) (G A1 + G A2) / (A _{01 G R1 + a 02 G R2} + a 03 G R3) = 21 / (21-O 2) (3) in addition, the description of the symbols in the above equation (1) to (3) are as follows. G _{R1, 2, 3:} the incinerated supply amount [kg / h] H _{U1, 2, 3:} lower heating value [kcal / kg] K _{1, 2, 3:} the layer in the combustion rate A _{01, 2, 3:} complexity theory air [Nm ³ / kg] C _B: layer material specific heat [kcal / kg ° C.] W _B: layer material weight (kg) C _PG: exhaust gas specific heat [kcal / Nm ³ ° C.] C _PA: air specific heat [kcal / Nm ³ ° C] G _A1 : Flowing air flow rate [Nm ³ / h] T _A1 : Flowing air temperature [° C] G _A2 : Secondary air flow rate [Nm ³ / h] T _A2 : Secondary air temperature [° C] G _SP : Water injection amount in bed [kg / h] Q _BI : Heat input of bed material [kcal / h] Q _BO : Heat output of bed material and incombustible material [kcal / h] _TG : Furnace outlet gas temperature [° C] T _B : _Bed temperature [° C] O ₂ : Furnace outlet oxygen concentration [%]

Bed temperature T _B of [0008] Formula (1) are that are controlled to be constant, since a large heat capacity of the layer material, the time change is gradual, the change in a short time is negligible can do. Therefore, when applying the method of the present invention, since it may be considered that dT _B / dt = 0, equation (1) to (3) is an unknown to be incinerated supply amount G _R1, G _R2, G _R3 It becomes a simultaneous algebraic equation and can be solved in real time using the neural network 28 shown in FIG. In the equations (1) to (3), the measured values are the flow air flow rate G _A1 , the flow air temperature T _A1 , the secondary air temperature G _A2 , the secondary air temperature T _A2 , the water injection amount G _SP in the bed, the furnace The outlet gas temperature T _G , the bed temperature T _B , and the oxygen concentration O ₂ at the furnace outlet. Since the other values are substantially constant, they can be treated as coefficients.

The neural network 28 shown in FIG. 2 will be described in more detail. Formulas (1) to (3) are formulas (4) and (4) in which the incineration material supply amounts are x ₁ , x ₂ , and x ₃ .
It can be transformed into simultaneous algebraic equations of the forms (5) and (6). _{a 11 x 1 + a 12 x} 2 + a 13 x 3 = b 1 (4) a 21 x 1 + a 22 x 2 + a 23 x 3 = b 2 (5) a 31 x 1 + a 32 x 2 + a 33 x 3 = b ₃ (6) Since this is a modification of equations (1) to (3), equation (4)
The meaning of each element of (5) and (6) is as follows. _{a 11 = H U1 K 1,} a 12 = H U2 K 2, a 13 = H U3 K 3 a 21 = H U1 (1-K 1), a 22 = H U2 (1-K 2), a
_{23 = H U3 (1-K} 3) a 31 = A 01, a 32 = A 02, a 33 = A 03 b 1 = -C PA G A1 T A1 -Q BI + Q BO + C PG T B {G A1 +
(22.4 / 18) G _SP ｝ +600 G _SP b ₂ = C _PG T _G ｛G _A1 + G _A2 + (22.4 / 18)
_{G SP} -C PG T B {} G A1 + (22.4 / 18) G SP} -
C _PA T _A2 G _A2 b ₃ = (G _A1 + G _A2 ) (21−O ₂ ) / 21 x ₁ = G _R1 , x ₂ = G _R2 , x ₃ = G _R3 Measured values are G _A1 , T _A1 , G _{A2, T A2, G SP,} T G, T B,
Since it is O ₂ , b ₁ , b ₂ , and b ₃ are time-varying data. These three are inputs to the neural network. a ₁₁ , a ₁₂ , a ₁₃ , a ₂₁ , a ₂₂ , a ₂₃ , a ₃₁ ,
a ₃₂ and a ₃₃ are time-invariant data and are set as coefficients inside the neural network. x ₁ , x ₂ , and x ₃ are the outputs of the neural network and are the solutions of the simultaneous algebraic equations. _{μ 11, μ 12, μ 13} , μ 21, μ 22, μ 23,
μ ₃₁ , μ ₃₂ , and μ ₃₃ are gains selected in consideration of a trade-off between convergence and stability of the neural network.
When the gain is increased, the output converges faster but the output may become unstable. When the gain is decreased, the output is stable but it takes time to converge.

In FIG. 2, numerals 30, 32, 34, 3
The symbol drawn as 6 is an arithmetic function for multiplying an input signal from the left by a constant and outputting it to the right. For example, if Y → a ₂₃ → Z, then Z = a ₂₃ Y Y → μ ₂₂ → Z, then Z = μ ₂₂ Y Y → −1 → Z, then Z = −Y. Further, in FIG. 2, a symbol drawn like numeral 38 is an arithmetic function for integrating an input signal from the left and outputting it to the right. For example, if Y → ∫ → Z, then Z = ∫Ydt + C. Note that t is time and C is an initial value. The neural network outputs x ₁ , x ₂ , x ₃ and simultaneously x
₁ , x ₂ and x ₃ are used for the calculation by feedback. By repeating this, the outputs x ₁ , x ₂ , and x ₃ converge and become the solutions of the equations (4), (5), and (6). Since the co-firing rate is a ratio of the supply amount of each incineration material, it can be obtained by using the derived x ₁ , x ₂ , and x ₃ . For example, co-firing rate of _{x 1: x 1 / (x} 1 + x 2 + x 3) x 1 and the sum of the mixed combustion ratio of _{x 2: (x 1 + x} 2) / (x 1 + x 2 + x 3)

FIG. 3 is an overall view of a control system including the neural network 28. The control system of the fluidized-bed incinerator 10 (for example, a fluidized-bed industrial incinerator) is more complicated than that of FIG. 3, and FIG. 3 shows only the signal flow related to the estimation of the co-combustion rate by the neural network 28. The computing unit 40 calculates the set values of the furnace outlet gas temperature T _G , the flow air flow rate G _A1 , and the secondary air flow rate G _A2 by using the estimated value of the co-firing rate. PID (proportional integ
ral and derivative) controller 42,
Reference numerals 44, 46, and 48 denote control devices for canceling the difference between the set value and the measurement signal, and perform a proportional operation, an integral operation, and a differential operation. In the fluidized bed waste incinerator 10, there are more input / output signals than shown in FIG. 3, but here only the manipulated variables and measurement signals relevant to the method of the present invention are shown. The T _B (layer temperature) set value is a constant value, and the layer temperature is controlled to be this value.

FIGS. 4 to 7 show examples of data in a real furnace (fluidized bed industrial waste incinerator) to which the method of the present invention is applied. In each of the four graphs, the horizontal axis represents time, which is data for 8 hours from 8:00 to 16:00. Each graph will be described in the order of FIGS. 4 to 7. FIG. 4 shows the change over time of the estimated supply amount [t / H] of the two types of incinerated materials 1 + 2, and relates to the sum (x ₁ + x ₂ ) of the outputs x ₁ and x ₂ of the neural network. That is, the supply amount G _R1 [t /
H] and the estimated value of the supply amount G _R2 [t / H] of the incinerated material. G _R1 is the supply amount of incineration material 1,
Supply amount of the G _R2 be incinerated 2, the supply amount of the incinerated 3 G _R3. FIG. 5 shows a time-dependent change in the actual results [t] of the incinerated materials 1 + 2, and shows the total incineration results of the incinerated materials 1 and 2 every hour. Figure 6 shows the time course of the supply amount estimation value of the incinerated 3 [t / H], the present invention relates to an output x ₃ of the neural network. That is, it is an estimated value of the supply amount G _R3 [t / H] of the incineration material 3. Fig. 7 shows the conveyor speed [%] of the incinerated material 3
Of time, that is, the speed of the conveyer that conveys the incinerated material 3 into the furnace. Comparing the graphs of FIG. 4 and FIG. 5 and the graphs of FIG. 6 and FIG.

[0013]

As described above, the present invention has the following effects. (1) Using the measurement signals such as temperature and air flow rate and the signal processing by the neural network, the real-time estimation of the co-firing rate of the incinerators with different properties is performed, and the estimated value of the co-firing rate is used to measure the furnace outlet gas temperature, Since the setting values of the flow air flow rate and the secondary air flow rate are derived, effective combustion control can be performed according to the supply amount of the incineration object, and CO, NOx
Can be suppressed.

[Brief description of the drawings]

FIG. 1 is a schematic structural view of a fluidized bed furnace used to carry out the method of the present invention.

FIG. 2 is an illustration of a neural network used to implement the method of the present invention.

FIG. 3 is a system diagram showing a co-firing rate estimation / combustion control system for implementing the method of the present invention.

FIG. 4 is a graph showing a change with time of an estimated supply amount of the incinerated material 1 + 2 in data of an actual furnace to which the method of the present invention is applied.

FIG. 5 is a graph showing the change over time in the actual performance of the incinerated material 1 + 2.

FIG. 6 is a graph showing a change with time of an estimated supply amount of the incineration material 3;

FIG. 7 is a graph showing a change over time in a conveyor speed of an incinerated material 3;

[Explanation of symbols]

 Reference Signs List 10 fluidized bed furnace (fluidized bed waste incinerator) 12 primary air 16 fluidized bed 18 freeboard unit 20 secondary air 22 cooling water 28 neural network 40 arithmetic unit 42 PID controller 44 PID controller 46 PID controller 48 PID control vessel

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location F23G 5/50 ZAB F23G 5/50 ZABM ZABN (72) Inventor Yuichi Miyamoto 1 Kawasaki-cho, Akashi-shi, Hyogo No. 1 Kawasaki Heavy Industries Ltd., Akashi Plant (72) Inventor Hidetaka Miyazaki 1-1 Kawasaki-cho, Akashi-shi, Hyogo Prefecture Kawasaki Heavy Industries Ltd., Akashi Plant (72) Inventor Masato Hayashi 1-1, Kawasaki-cho, Akashi-shi, Hyogo Prefecture Issue Kawasaki Heavy Industries, Ltd. Akashi Plant (72) Inventor Kaoru Oyano 1-1 Kawasaki-cho, Akashi-shi, Hyogo Prefecture Kawasaki Heavy Industries Ltd. Akashi Plant (72) Inventor Genichiro Nakanishi 1-3-1, Higashikawasaki-cho, Chuo-ku, Kobe No. Kawasaki Heavy Industries, Ltd. Kobe Head Office (72) Inventor Riki Koshida 1-3 1-3 Higashikawasaki-cho, Chuo-ku, Kobe-shi Kawasaki Heavy Industries Formula company Kobe in the head office

Claims (2)

[Claims] 1. When burning an incineration object in a fluidized bed incinerator, it is possible to estimate the co-firing rate of incineration objects having different properties in real time by using a signal processed by a fluidized bed incinerator and a signal processing by a neural network. By deriving the furnace outlet gas temperature, the flow air flow rate and the secondary air flow rate according to the supply amount of the incineration object, determining the manipulated value from these values, and performing combustion control. Method for estimating mixed combustion rate and controlling combustion in a furnace. 2. The measurement signal is a flow air flow, a flow air temperature, a secondary air flow, a secondary air temperature, a water injection amount in a bed, a furnace outlet gas temperature, a bed temperature, and a furnace outlet oxygen concentration. 2. The method for estimating and controlling the co-firing rate of a fluidized-bed incinerator according to 1.