JPH05317643A - Method for controlling circulating flow rate of liquid absorbent for wet flue gas desulfurizer and device therefor - Google Patents

Abstract

PURPOSE:To provide a device for controlling the circulating flow rate of a liquid absorbent for a wet flue gas desulfurizer where the circulating flow rate of the liquid absorbent is controlled to keep the desulfurization efficiency at the desired value without measuring the sulfite concentration in the liquid absorbent. CONSTITUTION:A exhaust gas flow rate 2, a liquid absorbent pH value 3, an inlet SO2 concentration 4 and a desulfurization rate 6 are inputted into an on-line measured value related estimator 36 consisting of a hierarchical neural net to calculate the liquid absorbent circulating flow rate, and a weight coefficient 27 of neurone bond in the estimator 36 is determined so that deviation between the liquid absorbent circulating flow rate and the actual circulating flow rate may not be more than the prescribed value. Next, an exhaust gas flow rate 7 after elapsing of time of dt, a liquid absorbent pH at 8 and an inlet SO2 concentration 9 and the neurone bond weight coefficient 27 in the estimator 36 are inputted into a flow rate demand preceded value calculator 37 having a neural net of the same constitution as the estimator 36 to obtain a liquid absorbent circulating flow rate preceded value 12, a signal 12 being corrected by a deviation signal 14 of a desulfurization rate 13 obtained from the inlet SO2 concentration 9 and an outlet SO2 concentration 10 and a desulfurization rate set value 11 to determine the number of liquid absorbent pumps in operation.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for controlling a circulating flow rate of an absorbent in a wet exhaust gas desulfurization apparatus, and in particular, appropriately controlling a circulating flow rate of an absorbing tower to reduce power of an absorbing tower circulating pump at a low load. The present invention relates to a method and an apparatus for controlling an absorption liquid circulation flow rate of a wet exhaust gas desulfurization apparatus suitable for

[0002]

2. Description of the Related Art A wet exhaust gas desulfurization apparatus, as shown in FIG. 6, brings an inlet exhaust gas 40 into a gas-liquid contact with an absorption liquid supplied from an absorption liquid circulation line 41 in an absorption tower 43, so that SO ₂ in the exhaust gas is Fixed in the form of sulfite in the absorbent,
The exhaust gas is discharged from the stack through the discharge line 44.
The absorption liquid that has absorbed SO ₂ flows down from the tower portion to the circulation tank 45. The circulation tank 45 is supplied with an absorbent such as limestone slurry through an absorbent slurry flow rate adjusting valve 46, and S
The liquid that has recovered the O ₂ absorption performance is supplied into the absorption tower 43 by the absorption tower circulation pump 50 and sprayed. Part of the circulating liquid is discharged through the extraction line 42, and in a subsequent step, the sulfite in the absorbing liquid is oxidized and recovered as gypsum.

A related control system of this type of wet flue gas desulfurization apparatus is, for example, JP-A-60-110320.
The gazette is cited. In this control method, the optimum pH value signal 4 of the absorption liquid circulating through the absorption tower is calculated by the simulation model 48 according to the load amount of the exhaust gas flowing into the absorption tower.
9 and the optimum operating number signal 47 of the absorption tower circulation pump 50
When the load is stable, the optimum operating number is subtracted from the set number by 1, and a certain increment is added to the above-mentioned optimum pH value to make it the set value of pH.
8, the absorbent supply amount and the number of pumps are controlled based on the changed set value only when the desulfurization rate satisfies the target value.

However, in this control method, it is essential that the simulation model can accurately reproduce the behavior of the actual machine. In the desulfurization device, the desulfurization performance is governed by the exhaust gas flow rate, the inlet SO ₂ concentration, the absorption liquid pH and the liquid-gas ratio, but even at the same pH, it is desulfurized depending on the oxidation state in the absorption liquid, that is, the concentration of sulfite. Performance is different.

FIG. 5 shows the relationship between the oxidation state and desulfurization performance.
As is clear from the figure, the oxidation state, that is, the concentration of sulfite, is required to accurately predict the change in the desulfurization rate due to the change in operating conditions by simulation, and this cannot be measured online. Considering the uncertainty of the oxidation rate, it is necessary to correct the data by manual analysis and it is complicated in driving operation.
Since the operator intervenes in this data correction work,
No consideration was given to the possibility of human error.

[0006]

In the above-mentioned prior art, the optimum operating number of absorption tower circulation pumps is determined by a simulation model. However, the accuracy of the simulation model is not taken into consideration, and it is necessary if the accuracy decreases. Problem that the desulfurization rate cannot be maintained, and when the operating conditions change drastically, it is necessary to correct the coefficients of the simulation model etc. by the manual analysis value of the liquid composition, which increases the burden on the operator. was there.

On the other hand, there is a method of applying fuzzy control as a control device that can maintain the desulfurization rate near the target value using only the information that can be measured online, but even in this case, the information collected through interviews with skilled operators is summarized. Then, there is a problem that the control system designer has to formalize as a control rule and is not necessarily an optimum control system.

An object of the present invention is to eliminate the above-mentioned problems of the prior art, to use only the on-line measured amount, and to maintain the desulfurization rate near the target value without presetting control rules or the like. It is an object of the present invention to provide a method and an apparatus for controlling an absorption liquid circulation flow rate of an apparatus.

[0009]

To achieve the above object, the first invention of the present application is based on the exhaust gas flow rate and SO ₂ concentration at the desulfurization unit inlet, the absorption liquid pH value of the unit and the desulfurization rate. In the method for controlling the absorption liquid circulation flow rate in the wet exhaust gas desulfurization device, the treated exhaust gas flow amount, the inlet exhaust gas SO ₂ concentration, the desulfurization rate, and the absorption liquid pH value, which are obtained in advance by measurement in the desulfurization device in operation, are hierarchically calculated. It is input to the neural network to calculate the absorption tower circulation flow rate in the desulfurizer, and the weight of the connection between the neurons of the neural network is adjusted so that the difference between the calculated value and the actual absorption circulation flow rate is within a predetermined range. leave determine the coefficients, treated flue gas flow rate at the desulfurization apparatus after a predetermined time, the measured value and the desulfurization ratio setting value of the inlet flue gas SO ₂ concentration and the absorption liquid pH value, the said hierarchical neural network A wet type characterized by calculating a required absorption liquid circulation flow rate of the absorption tower by using a weighting factor of the connection between the determined neurons and inputting it to the configured neural net, and controlling the flow rate based on this. The present invention relates to a method for controlling an absorption liquid circulation flow rate of an exhaust gas desulfurization device.

A second aspect of the present invention is a method for controlling the circulating flow rate of an absorbent in a wet exhaust gas desulfurization apparatus based on the exhaust gas flow rate and SO ₂ concentration at the desulfurization apparatus inlet and the absorption solution pH value and desulfurization rate of the apparatus. In, the treated exhaust gas flow rate, the inlet exhaust gas S, which was previously obtained by measurement with the desulfurization apparatus in operation
The O ₂ concentration, the desulfurization rate, and the absorption liquid pH value are input to the hierarchical neural net to calculate the absorption tower circulation liquid of the absorption tower in the desulfurization device, and the difference between the calculated value and the actual absorption liquid circulation flow is predetermined. The weighting coefficient of the coupling between the neurons of the neural network is determined so as to be within the range, and the measured exhaust gas flow rate, the inlet exhaust gas SO ₂ concentration, and the absorption liquid pH value in the desulfurization device after a predetermined time and the desulfurization The desulfurization rate and desulfurization rate set value calculated from the measured SO ₂ concentration at the desulfurization apparatus inlet and outlet by calculating the required absorption liquid circulation flow rate of the absorption tower by inputting the rate setting value into the hierarchical neural network. A method for controlling an absorption liquid circulation flow rate of a wet exhaust gas desulfurization apparatus, characterized in that the required absorption liquid circulation flow rate calculated value is corrected based on the deviation amount and the absorption tower absorption liquid circulation flow rate is controlled by the corrected flow rate.

A third aspect of the present invention is a wet type exhaust gas desulfurization for controlling the absorption liquid circulating flow rate in the absorption tower based on the exhaust gas flow rate and SO ₂ concentration at the desulfurization device inlet, the desulfurization device absorption liquid pH value, and the desulfurization rate. In the controller of the system, the treated exhaust gas flow rate, inlet exhaust gas SO ₂ concentration, absorption liquid pH value, and desulfurization rate obtained by measurement in the desulfurization device are input to calculate the absorption liquid circulation flow rate in the absorption tower, and the calculated value So that the difference between the actual circulation flow rate and the actual circulation flow rate is within a predetermined value, an on-line measurement quantity relation estimation device that corrects and determines the connection weight coefficient between the neurons of the hierarchical neural network therein, and the desulfurization device after a predetermined time treated flue gas flow rate, inlet gas SO ₂ density, absorption liquid pH value and using a weighting factor binding between the desulfurization ratio setting value and the online measuring relationship neurons determined by estimating apparatus in A flow rate demand prior value calculator configured in a hierarchical neural network for calculating the absorption liquid circulation flow rate of the absorption tower Te, SO ₂ in the desulfurization apparatus inlet and outlet
Absorption based on the output signals of the flow rate advance value correction device that generates a correction signal based on the deviation between the desulfurization rate and the desulfurization rate set value obtained from the concentration measurement, and the output signals of the flow rate demand advance value calculator and the flow rate advance value correction device. The present invention relates to an absorbent circulating flow rate control device for a wet exhaust gas desulfurization apparatus, which is provided with means for controlling a tower absorbent circulating flow rate.

A fourth invention is the same as the third invention,
Wet exhaust gas desulfurization apparatus characterized by including means for determining the number of operating absorption tower circulation pumps based on the absorption tower circulation flow rate of the absorption tower determined by the output signals of the flow rate demand advance value calculator and the flow rate advance value correction device. Of the absorption liquid circulation flow rate control device.

[0013]

[Advantage] Since the absorption liquid circulation flow rate preceding value based on the online measurement amount is a signal calculated while estimating the relational expression established between the online measurement amounts of the desulfurization system momentarily, the desulfurization rate is set to the set value. It operates so as to adapt the flow rate demand, which is the basis for maintaining it, in accordance with changes in operating conditions. As described above, the actual desulfurization rate deviation is made to approach zero by the flow rate demand that is adaptively calculated with respect to the change in the operating state, so that the desulfurization rate does not deviate from the target.

[0014]

FIG. 1 shows a concrete example of a method for controlling a circulation flow rate of an absorption tower of a wet flue gas desulfurization apparatus according to the present invention, to which a neural net is applied. In the figure, 37 is a flow rate demand advance value calculator, and at a certain time t, the exhaust gas flow meter 2
2, the pH meter 23, the inlet SO ₂ concentration meter 24, the exhaust gas flow rate 7 at the time t which is the output signal of the desulfurization rate setting device 26, the pH at the time t 8, the inlet SO ₂ concentration 9 at the time t,
The desulfurization rate set value 11 is used to calculate and calculate the absorption liquid circulation flow rate 12. This flow rate demand advance value calculator 37 has an exhaust gas flow rate 7 at time t, a pH 8 at time t, and an inlet S at time t.
The O ₂ concentration 9 and desulfurization rate set value are used as input signals, and the absorption liquid circulation flow rate preceding value 12 is output.
Is the exhaust gas flow rate 2 at the time point t-dt, the pH 3 at the time point t-dt, the inlet SO ₂ concentration 4 at the time point t-dt, and the desulfurization rate 6 at the time point t-dt. The weight coefficient 27 is set to be equal to that between the neurons constituting the neural network that outputs the absorbing liquid circulation flow rate 1 at the time point t-dt, that is, the online measurement amount relationship estimator 36.

On the other hand, the inlet SO ₂ concentration meter 24 and the outlet SO ₂ concentration meter
From the inlet SO ₂ concentration 4 and the outlet SO ₂ concentration 10 at the time t which is the output signal of the densitometer 25, the desulfurization rate 13 at the time t is obtained by using the subtractor 28a and the divider 29a, and the subtraction is performed. Bowl 2
In 8b, the desulfurization rate deviation signal 14 with respect to the desulfurization rate set value 11 is obtained and input to the PI controller 15. The PI controller 15 outputs the flow rate advance value correction signal 16 from the desulfurization rate deviation signal 14 by the PI operation, and the adder 38 adds it to the absorbing liquid circulation flow rate advance value 12 to calculate the flow rate demand signal 17, which is then pumped. The circulating flow rate is controlled by inputting to the number setting device 18 to calculate the optimum operating number signal 19 and finally setting the operating number of the absorption tower circulation pump 20.

The internal structure of the online measurement quantity relationship estimator 36 will be described below with reference to FIG. The on-line measured amount relation estimator 36 uses the exhaust gas flow rate 2 at the time point t-dt,
PH 3 at time t-dt, inlet SO ₂ at time t-dt
Desulfurization rate 6 at concentration 4 and time t-dt
1a, 31b, 31c and 31d, respectively, and outputs 33a, 33b, 33c and 33 from the respective neurons.
d and a constant value output 33e (-1) from the Shiki value neuron 32a are multiplied by the weight of the connection from each neuron to the input target neuron 31f by the weighted adder 34a, and then added. , The input 35a to the neuron 31f is calculated. Here, the neuron 31f is an arbitrary neuron in the second layer. That is, the inputs of the neurons 31a, 31b, 31c, 31d are X1 = Gg (t-dt), X2 = pH (t-dt), X3 = SO ₂ in (t-dt), X4 = ETA (t-dt), Then, the neurons 31a, 31b, 31c, 31d
Outputs these as they are. That is, the outputs 33a, 3
3b, 33c, and 33d are none other than the above signals. Here, Gg (t-dt), pH (t-dt), S in the above formula.
O ₂ in (t-dt) and ETA (t-dt) represent the exhaust gas flow rate, pH, inlet SO ₂ concentration, and desulfurization rate as a function of time, respectively.

Next, the i-th neuron 31f in the second layer
The input 35a of is given by the following equation. X2i = W12i * X1 + W12i * X2 + W13i *
X3 + W14i * X4-W15i The neuron 31f processes this input X2i according to the following relation to calculate the output 33f. That is, the output 33f
Is the next signal.

Y2i = sigm (X2i) Here, the function sigm (X) is defined by sigm (X) = tanh (X / 2) and is called a sigmoid function. These second
Layer neurons 31e, 31f. ． ． The output from is input to the weighted adder 34b again, and the weighted adder 34b
From b: X33 = W231 * Y21 + W232 * Y22 +. ． ．
+ W23i * Y2i +. ． ． + W23n * Y2n-W2
3 is added. Here, n is the number of neurons in the second layer, and W23l ,. ． ． W23i ,. ． ． W23n, W2
3 is a weighting coefficient. This signal X33 is input to the neuron 31z of the third layer and also processed and output by the sigmoid function.

Y3 = sigm (X33) This output Y3 is compared with the absorption liquid circulation flow rate l at the time point t-dt, which is designated as L (t-dt), which is Y3 in the error calculator 30. And L (t-dt), the square of the difference is calculated, and the weight coefficient W12i ,. ． ． , W23
1 ,. ． ． , W23 is determined.

The flow rate demand advance value calculator 37 is a neural network having the same structure as the online measurement quantity relationship estimator, which will be described with reference to FIG. The flow rate demand advance value calculator 37 indicates that the exhaust gas 7 at the time point t, the pH 8 at the time point t, the time point t
The inlet SO ₂ concentration 9 and the desulfurization rate set value 11 at are input to the neurons 31a ′, 31b ′, 31c ′, 31d ′ respectively, and the outputs 33a ′, 33 from the neurons are input.
b ′, 33c ′, 33d ′ and Shiki value neuron 3
A constant value output -1 from 2a 'is input to each neuron 31f by a weighted adder.
Calculate input 35a 'for'. Here, the neuron 31f 'is an arbitrary neuron in the second layer. That is, the neurons 31a ', 31b', 31c ', 31d
When the input of ′ is X1 ′ = Gg (t) X2 ′ = pH (t) X3 ′ = SO ₂ in (t) X4 ′ = ETASET, the neurons 31a ′, 31b ′, 31c ′,
31d 'outputs these as they are. Ie output 3
3a ', 33b', 33c 'and 33d' are none other than the above signals. Next, the i-th neuron 31f in the second layer
The input 35a 'of' is given by the following equation.

X2i '= W12i * X1' +. ． ． + W
14i * X4'-W15i, where the weights W12i ,. ． ． , W14i, and W15i are neural nets that constitute the online measurement quantity relationship estimator 36, and values determined by the back propagation method, for example, are applied. The neuron 31f 'processes this input X2i' according to the following relation to calculate the output 33f '. That is, the output 33f 'is the next signal.

Y2i '= sigm (X2i') Further, these second layer neurons 31e ', 31f
´. ． ． The output from is again input to the weighted adder 34b ', and X33' = W231 * Y21 '+ W232 * Y22' in the weighted adder 34b '.
+. ． ． + W23n * Y2n'-W23 is calculated. Here, W231 ,. ． ． ． ． W23 is a neural network which is an internal configuration of the online measurement quantity relationship estimator 36 and applies a value determined by, for example, the back propagation method. This signal X33 'is the third layer neuron 31z'
Is input to and is also processed by the sigmoid function and output.

Y3 '= sigm (X33') Then, this output is output from the flow rate demand preceding value calculator 37 as the absorbing liquid circulation flow rate preceding value 12. This control system basically calculates the flow rate demand by adding the absorption liquid circulation flow rate preceding value 12 and the desulfurization rate deviation signal 14 to the flow rate preceding value correction signal 16 which is feedback-corrected by the PI controller 15. The characteristic is that the relationship between the online measurement amounts is estimated by a neural network, and the absorption liquid circulation flow rate preceding value 12 is calculated back from the desulfurization rate set value using this estimation result.

As shown in FIG. 5, in the desulfurization apparatus, the desulfurization rate varies greatly depending on the concentration of sulfite produced in the absorption liquid by absorbing SO ₂ in the exhaust gas, that is, the oxidation state. .. Therefore, even if the same desulfurization rate deviation occurs, the absorption tower circulation flow rate value that must be increased or decreased to maintain the desulfurization rate at the target value varies depending on the oxidation state. That is, the feedback correction is performed by the normal P
When the I controller is used, the proportional gain and the integration time are usually constant, so that the adaptive correction due to the difference in the oxidation state as described above is impossible.

In the present invention, the flow rate demand advance value calculator 3
7, the flow rate demand advance value is calculated based on the estimation of the relationship between the process variables by the neural network as shown in FIG. 4, so that the desulfurization rate is maintained near the target value in all operating conditions. As described above, in the present invention, the preceding flow rate value operates as if controlled by a skilled operator, so that the desulfurization rate can be controlled to be close to the target value even in a special operating condition. In the description of the embodiment using FIGS. 1, 2, and 3, the online measurement quantity relation estimator 36 and the flow rate demand preceding value calculator 37 are separate neuron networks.
And 37 as the same neuron network,
It is also possible to implement the present invention by switching the input signal entering the estimator 36 at time t-dt and the input signal entering the calculator 37 at time t using a switch. Included in the present invention.

[0026]

According to the present invention, a flow rate demand advance value calculator to which a neural network is applied is installed to calculate the absorption liquid circulation flow rate advance value. This flow rate demand advance value calculator is Since the structure has the same weight coupling as the neural network of the online measurement quantity estimator, which estimates the relationship between the online measurement quantities moment by moment, the flow rate demand can be changed momentarily while observing the change behavior of the desulfurization rate. The preceding value is given to determine the operating number of the absorption tower circulation pump, and it becomes possible to control the absorption tower circulation flow rate as if by an experienced operator who is familiar with the behavior of the plant.

Therefore, stable desulfurization performance is ensured even when the oxidation state of the absorbing liquid changes not only under normal operating conditions, but the relationship between the online measurement amounts of the desulfurization system changes significantly. Therefore, there is an effect that the absorption tower circulation pump power at a low load can be reduced by appropriately controlling the absorption tower circulation flow rate.

[Brief description of drawings]

FIG. 1 is a control system diagram showing an embodiment of an absorption tower circulation flow rate control method according to the present invention.

FIG. 2 is an explanatory diagram showing an embodiment of a neural network which is an internal structure of an online measurement quantity relationship estimator.

FIG. 3 is an explanatory diagram showing an embodiment of a neural network which is an internal structure of a flow rate demand advance value calculator.

FIG. 4 is a principle diagram for setting the number of circulation pumps.

FIG. 5 is an explanatory view showing the relationship between pH and oxidation state and desulfurization rate.

FIG. 6 is a control system diagram showing a conventional absorption tower circulation flow rate control method.

[Explanation of symbols]

 1 ... Absorption liquid circulation flow rate at time point (t-dt), 2 ... time point
Exhaust gas flow rate at (t-dt), 3 ... At time (t-dt)
PH, 4 ... inlet SO at time point (t-dt)₂Concentration, 5
... Exit SO at time point (t-dt)₂Concentration, 6 ... Time (t
Desulfurization rate at -dt), 7 ... Exhaust gas flow rate at time t, 8 ...
PH at time t, 9 ... inlet SO at time t ₂Concentration 10
… Exit SO at time t₂Concentration, 11 ... Desulfurization rate set value, 1
2 ... Leading value of circulating flow rate of absorbing liquid, 13 ... Desulfurization rate at time t,
14 ... Desulfurization rate deviation signal, 15 ... PI controller, 16 ... Flow rate
Leading value correction signal, 17 ... Flow rate demand signal, 18 ... Pon
Number setting device, 19 ... Optimal operating signal, 20 ... Absorption tower
Circulation pump, 21 ... Circulation flow meter, 22 ... Exhaust gas flow rate
Total, 23 ... pH meter, 24 ... Inlet SO₂Densitometer, 25 ...
Mouth SO₂Densitometer, 26 ... Desulfurization rate setting device, 27 ... Weighting factor
Signal, 28 ... Subtractor, 29 ... Divider, 30 ... Error calculation
Vessel, 31 ... Neuron, 32 ... Shiki value neuron, 3
3 ... Output signal, 34 ... Weighted adder, 35 ... Input signal
No., 36 ... Estimator for online measurement relation, 37 ... Flow rate data
Mand leading value calculator, 38 ... Adder.

Claims (4)

[Claims] 1. Exhaust gas flow rate and SO at the desulfurizer inlet
₂ Concentration, in the method of controlling the absorption liquid circulation flow rate in the wet exhaust gas desulfurization device based on the absorption liquid pH value and desulfurization rate of the device, in advance, the treated exhaust gas flow rate obtained by measurement in the desulfurization device in operation The inlet exhaust gas SO ₂ concentration, the desulfurization rate, and the absorption liquid pH value are input to a hierarchical neural net to calculate the absorption tower circulation liquid flow rate in the desulfurization device,
The weighting coefficient of the coupling between the neurons of the neural network is determined in advance so that the difference between this calculated value and the actual circulating flow rate of the absorbing liquid is within a predetermined range, and the treated exhaust gas flow rate at the desulfurization unit after a predetermined time, the inlet The measured values of the exhaust gas SO ₂ concentration and the absorption liquid pH value and the desulfurization rate set value are input to a neural network having the same structure as the hierarchical neural network, and the weighting coefficient of the connection between the determined neurons is used. A method for controlling a circulating flow rate of an absorbent for a wet exhaust gas desulfurization apparatus, which comprises calculating a required circulating flow rate of an absorbent in an absorption tower and controlling the flow based on the calculated flow rate. 2. Exhaust gas flow rate and SO at the desulfurizer inlet
₂ Concentration, in the method of controlling the absorption liquid circulation flow rate in the wet exhaust gas desulfurization device based on the absorption liquid pH value and desulfurization rate of the device, in advance, the treated exhaust gas flow rate obtained by measurement in the desulfurization device in operation The inlet exhaust gas SO ₂ concentration, the desulfurization rate, and the absorption liquid pH value are input to a hierarchical neural net to calculate the absorption tower circulation liquid flow rate in the desulfurization device,
The weighting coefficient of the coupling between the neurons of the neural network is determined in advance so that the difference between this calculated value and the actual circulating flow rate of the absorbing liquid is within a predetermined range, and the treated exhaust gas flow rate at the desulfurization unit after a predetermined time, the inlet The measured values of exhaust gas SO ₂ concentration and absorption liquid pH value and desulfurization rate set value are input to the hierarchical neural network to calculate the required absorption liquid circulation flow rate of the absorption tower, and the SO ₂ concentration at the desulfurization device inlet and outlet. The required absorption liquid circulation flow rate calculation value is corrected based on the deviation amount between the desulfurization rate and the desulfurization rate set value obtained from the measured value, and the absorption tower absorption liquid circulation flow rate is controlled by the corrected flow rate. Method for controlling a circulating flow rate of an absorbent in a wet exhaust gas desulfurization apparatus. 3. Exhaust gas flow rate and SO at the desulfurizer inlet
₂ concentration, desulfurizer absorption liquid pH value, in the control device of the wet exhaust gas desulfurization device to control the absorption liquid circulation flow rate in the absorption tower based on the desulfurization rate, the treated exhaust gas flow rate obtained by measurement in the desulfurization device, Inlet exhaust gas SO ₂ concentration, absorption liquid p
The H value and the desulfurization rate are input to calculate the circulating flow rate of the absorbing liquid in the absorption tower, and the neurons of the hierarchical neural network in the inside are adjusted so that the difference between this calculated value and the actual circulating flow rate is within a predetermined value. An on-line measurement amount relation estimation device for correcting and determining the coupling weight coefficient between the _two , a treated exhaust gas flow rate in the desulfurization device after a predetermined time, an inlet exhaust gas SO ₂ concentration, and an absorbent p
A flow rate demand composed of a hierarchical neural network for calculating the absorption liquid circulation flow rate in the absorption tower using the H value, the desulfurization rate set value, and the weighting factor of the coupling between the neurons determined by the online measurement relation estimation device. A preceding value calculator, a preceding flow value correction device that generates a correction signal based on a deviation between the desulfurization rate and a desulfurization rate set value obtained by measuring the SO ₂ concentration at the inlet and outlet of the desulfurization device, and the preceding flow rate demand preceding value calculation And a means for controlling the absorption tower circulation flow rate of the absorption tower based on the respective output signals of the tank and the flow rate preceding value correction apparatus. 4. The means for determining the number of operating absorption tower circulation pumps according to claim 3, based on the absorption tower circulation liquid circulation flow rate obtained from the output signals of the flow rate advance value calculator and the flow rate advance value correction device. An absorption liquid circulation flow rate control device for a wet exhaust gas desulfurization device.