CN113485106A - Method for controlling concentration of nitrogen oxide in thermal power plant - Google Patents

Abstract

The invention provides a method for controlling the concentration of nitrogen oxides in a thermal power plant, which specifically comprises the following steps: acquiring the operation data of the #1 unit from the field DCS and sequentially carrying out normalization processing, filtering processing and zero initial value processing; respectively identifying a transfer function model of the process from the urea solution flow to the nitrogen oxide emission concentration of the SNCR denitration control system under typical working conditions of 170MW and 260MW by using an IPSO algorithm; on the basis of the original single-loop PID control strategy of a power station site, a method (AGA-Smith) combining an adaptive genetic algorithm and a Smith prediction compensation control method is introduced. Matlab simulation results show that under two typical working conditions, the AGA-Smith prediction compensation control overshoot is smaller, the external disturbance resistance is stronger, the model adaptation capability is stronger than that of single-loop PID control, good technical reference is provided for on-site SNCR denitration control of a power station, and the current situations that the accuracy of the SNCR denitration system model of a thermal power plant is not high and the denitration control effect is not good can be solved.

Description

Method for controlling concentration of nitrogen oxide in thermal power plant

Technical Field

The invention relates to the field of thermal power generation, in particular to a method for controlling the concentration of nitrogen oxide in a thermal power plant.

Background

The urea solution flow regulating loop controlled device of the power station on-site SNCR denitration control system is a urea solution main pipe electric regulating valve, the urea solution flow control adopts a mode of combining single-loop closed-loop control and table look-up, the injection amount of the urea solution corresponding to different loads or smoke gas amounts can be inquired in the table, the value found in the table is used as a feedforward value of the single-loop closed-loop control, and meanwhile, NO measured by the smoke gas on-line monitoring system is used as a feedforward value of the single-loop closed-loop control_xThe concentration feedback value corrects the flow of the sprayed urea solution.

The existing urea flow regulation control strategy does not consider the characteristics of large delay, large inertia and the like of a controlled object, the control effect is easy to generate large overshoot, and meanwhile, the existing urea flow regulation control strategy is used for ensuring NO at an outlet_xThe concentration reaches the standard in real time, and the direct result is that the injection amount of the urea solution is excessive, so that ammonia escapes, and the escaped ammonia forms ammonium salt and is adsorbed in a tail flue of a boiler, so that the corrosion of equipment is aggravated, and the economic benefit is influenced. Therefore, how to effectively overcome the large delay and large inertia characteristics of the controlled object is a key point for controlling the concentration of nitrogen oxides in the SNCR denitration control system.

Disclosure of Invention

The invention aims to overcome the defects in the prior art and provides a method for controlling the concentration of nitrogen oxide in a thermal power plant.

In order to realize the purpose, the invention is realized by the following technical scheme:

a method for controlling the concentration of nitrogen oxides in a thermal power plant comprises the following steps:

step 1: collecting urea solution flow x of unit under 170MW and 260MW working conditions_iWith nitrogen oxide concentration y_iData;

step 2: the flow rate x of the urea solution_iWith nitrogen oxide concentration y_iCarrying out normalization processing, filtering processing and zero initial value processing on the data in sequence;

and step 3: identifying parameters in the SNCR denitration control system model by using an IPSO algorithm based on the data processed in the step 2 to obtain a transfer function model from urea solution flow to nitric oxide concentration under the working conditions of 170MW and 260 MW;

and 4, step 4: and (3) on the basis of single-loop PID control, a method of combining an adaptive genetic algorithm with a Smith prediction compensation control method is introduced to control the transfer function model obtained in the step (3).

Further, the normalization processing in step 2 adopts the following formula:

wherein x is_iIs the flow rate of urea solution, x_min、x_maxMinimum and maximum values of the urea solution flow, x, respectively_i1The value is obtained after the urea solution flow is normalized; y is_iIs the concentration of nitrogen oxides, y_min、y_maxMinimum and maximum nitrogen oxide concentration, y_i1The nitrogen oxide concentration is normalized; 1,2,3.

The filtering processing in the step 2 adopts the following formula:

x_i2＝α₁x_i1+(1-α)x_i1′

wherein the smoothing coefficient alpha₁∈[0,1]；x_i1Is' x_i1Predicted value of (a), x_i2The value is obtained after filtering the flow of the urea solution; y is_i2The nitrogen oxide concentration is obtained by filtering; 1,2,3.

The zero initial value processing in the step 2 adopts the following formula:

wherein x is_i3The flow rate of the urea solution is obtained after zero initial value processing; y is_i3The nitrogen oxide concentration is obtained after zero initial value processing; 1,2,3.. 1800; n is 5.

Further, the model of the SNCR denitration system in step 3 is:

wherein K is a system open loop gain coefficient; t is₁.....T_nIs the inertia time constant; n is the system order; τ is the pure lag time constant.

Further, the specific process of identifying the parameters in the SNCR denitration control system model by the IPSO algorithm in the step 3 is as follows:

step 3.1: parameter initialization in IPSO algorithm, including population size D₁Number of iterations N₁Velocity v interval, inertia weight ω interval, learning factor C₁Learning factor C₂And the value ranges of an open loop gain coefficient K, an inertia time constant T, a system order n and a pure lag time constant tau in the SNCR denitration control system model;

step 3.2: calculating the fitness value of each particle in the initial population by taking the derivative of the mean square error function as a fitness function, and finding out the individual optimal value and the population optimal value in the initial population;

step 3.3: comparing the individual optimal value and the population optimal value in the initial population;

step 3.4: updating the position information and the speed information of the particles to obtain a new population;

step 3.5: calculating and comparing the individual optimal value and the group optimal value in the new population again;

step 3.6: when the maximum iteration number is met, optimizing is finished, K, T, n and tau after optimization are output, and otherwise, steps 3.2 to 3.6 are repeated.

Further, the population in step 3.1Scale D₁100, iteration number N ₁200, the speed interval v ∈ [ -1,1]The inertia weight interval ω ∈ [0.1,0.9 ]]Learning factor C₁＝C₂2.02, the open loop gain factor K e-5, 5]The inertia time constant T ∈ [1,400 ]]The system order n ∈ [1,4 ]]Pure lag time constant τ e [0,150 ∈ ]]。

Further, the speed information in step 3.4 is iteratively updated according to the following formula:

V_ij(t+Δt)＝ωV_ij(t)+C₁R₁[X_bestij-X_ij(t)]+C₂R₂[X_bestgj-X_ij(t)]

in the formula X_bestijIs the optimum position of the ith row and jth column of particles, X_bestgjThe optimal position of the group is obtained; c₁And C₂Is a learning factor; t is the current time, and t plus delta t is the time after delta t at the time t; r₁And R₂Is a random factor, R₁,R₂∈[0,1]；X_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.. said., 100, j 1,2, a.. said., 4; ω is the inertial weight, and the expression is as follows:

wherein ω is_maxIs the maximum inertia weight, omega_minIs the minimum inertia weight, k₁For the current number of iterations, k_1maxIs the maximum number of iterations.

The optimal position of the particles is determined by:

in the formula X_bestiIs the optimal position of the particles; q_bestiAn optimal fitness value for the particle; t is the current time, and t plus delta t is the time after delta t at the time t; x_iThe position of the ith particle.

The position information is iteratively updated according to the following formula:

X_ij(t+Δt)＝X_ij(t)+V_ij(t+Δt)

wherein t is the current time, and t plus delta t is the time after delta t at the time t; x_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.

Further, the transfer function model of the urea solution flow to the nitrogen oxide concentration under the 170MW condition in step 3 is:

the model of the transfer function from the urea solution flow to the nitrogen oxide concentration under the 260MW working condition is as follows:

further, the adaptive genetic algorithm in step 4 specifically includes the following steps:

step 4.1: initializing population, including K in PID controller_p、K_i、K_dValue ranges of three parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.2: coding the parameters in the step 4.1 by adopting a real number coding and linear interpolation mode;

step 4.3: selecting a fitness function Q (x) and calculating the fitness value of each individual;

step 4.4: selecting individuals meeting preset relative fitness value indexes by adopting a roulette method;

step 4.5: and (4) performing genetic operation on the individuals selected in the step 4.4, wherein the genetic operation formula is as follows:

wherein A, B is the individual selected by roulette method, A ', B' are the new individuals generated by genetic manipulation, α, β are constants, α, β ∈ [0,1 ].

Step 4.6: carrying out mutation operation on the individuals generated in the steps 4.4 and 4.5 by adopting a reverse method;

step 4.7: obtaining a new population after selection, heredity and mutation operations;

step 4.8: parameter decoding of new population, including K in PID controller_p、K_i、K_dThree parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.9: outputting the optimized K when the maximum iteration number is satisfied_p、K_i、K_dOtherwise, repeating steps 4.4 to 4.9.

Further, K in said step 4.1_p∈[-8,8]，K_i∈[1,10]，K_d∈[0,100]Population size D ₂50, iteration number N₂100, chromosome length L20, crossover probability P_c∈[0.4,0.9]The expression is as follows:

in the formula P_cmax，P_cminRespectively the maximum value and the minimum value of the cross probability; k is a radical of₂For the current number of iterations, k_2maxIs the maximum number of iterations.

Probability of variation P_mThe expression is as follows:

P_m＝0.2-[1:N₂]*0.04/N₂

the fitness function q (x) in step 4.3 has the following expression:

in the formula of omega₁、ω₂、ω₃Is constant, and ω₁+ω₂+ω₃1, e (t) is the deviation, t_sTo adjust the time, σ is the overshoot.

Further, the Smith estimation compensation control method in the step 4 is to connect the Smith estimation compensator in parallel on the PID controller in the reverse direction on the basis of the original single-loop PID control loop, and to use the SNCR denitration control system model as the model of the Smith estimation compensator.

The invention optimizes the K in the self-adaptive genetic algorithm_p、K_i、K_dThe three parameters are used as the three parameters of the PID controller, and then are combined with a Smith estimation compensation control method, so that the output of the concentration of the nitrogen oxide in the control model of the SNCR denitration system is controlled.

Compared with the prior art, the invention has the following beneficial effects:

(1) smith estimation compensation control is introduced, and the problems caused by large delay and large inertia of an SNCR denitration system are solved.

(2) Introducing self-adaptive genetic algorithm to improve the parameter K of the controller in the existing single-loop PID control_p、K_i、K_dThe parameter setting usually depends on the current situation of expert field adjustment and continuous debugging, and the optimal parameter is obtained by optimizing in a given parameter range through a self-adaptive genetic algorithm, so that the manual continuous debugging process is reduced.

(3) The self-adaptive genetic algorithm is combined with Smith prediction compensation control, the control effect on outlet nitrogen oxides is better, the overshoot index is smaller, the actual on-site is reflected, the usage amount of the urea solution is reduced, the ammonia escape phenomenon is reduced, the adverse problems caused by the ammonia escape phenomenon are reduced, and the economic benefit of a power plant is improved.

Drawings

FIG. 1 is a block diagram of IPSO algorithm optimization parameters;

FIG. 2 is a block diagram of an adaptive genetic algorithm;

FIG. 3 is a block diagram of AGA-Smith prediction compensation control;

FIG. 4 is a graph of identification effect under 170MW condition;

FIG. 5 is a graph showing the recognition effect under the 260MW condition;

FIG. 6 is a graph of control performance under 170MW condition;

FIG. 7 is a graph of control performance under 260MW condition;

FIG. 8 is a control performance curve under 170MW with K varied;

FIG. 9 is a control performance curve under 170MW when τ is changed;

FIG. 10 is a graph of control performance for 170MW with T varied.

Detailed Description

The following examples are given in the detailed description and the specific operation on the premise of the technical solutions of the present invention, but do not limit the protection scope of the patent of the present invention, and all technical solutions obtained by using equivalent alternatives or equivalent variations should fall within the protection scope of the present invention.

Examples

A method for controlling the concentration of nitrogen oxides in a thermal power plant comprises the following steps:

step 1: collecting urea solution flow x of #1 unit under typical working conditions of 170MW and 260MW from DCS system_iWith nitrogen oxide concentration y_iData;

step 2: the flow rate x of the urea solution_iWith nitrogen oxide concentration y_iCarrying out normalization processing, filtering processing and zero initial value processing on the data in sequence;

and step 3: based on the data processed in the step 2, identifying a transfer function model from the urea solution flow to the nitrogen oxide concentration of the SNCR denitration control system under the typical working conditions of 170MW and 260MW by using an IPSO algorithm;

and 4, step 4: and (3) on the basis of the existing single-loop PID control on the spot, a method of combining an adaptive genetic algorithm with a Smith prediction compensation control method is introduced to control the transfer function model obtained in the step (3).

Further, the normalization processing in step 2 adopts the following formula:

wherein x is_iIs the flow rate of urea solution, x_min、x_maxMinimum and maximum values of the urea solution flow, x, respectively_i1The value is obtained after the urea solution flow is normalized; y is_iIs the concentration of nitrogen oxides, y_min、y_maxMinimum and maximum nitrogen oxide concentration, y_i1The nitrogen oxide concentration is normalized; 1,2,3.

The filtering processing in the step 2 adopts the following formula:

x_i2＝α₁x_i1+(1-α)x_i1′

wherein the smoothing coefficient alpha₁∈[0,1]；x_i1Is' x_i1Predicted value of (a), x_i2The value is obtained after filtering the flow of the urea solution; y is_i2The nitrogen oxide concentration is obtained by filtering; 1,2,3.

The zero initial value processing in the step 2 adopts the following formula:

wherein x is_i3The flow rate of the urea solution is obtained after zero initial value processing; y is_i3The nitrogen oxide concentration is obtained after zero initial value processing; 1,2,3.. 1800; n is 5.

Further, the model of the SNCR denitration system in step 3 is:

wherein K is a system open loop gain coefficient; t is₁.....T_nIs the inertia time constant; n is the system order; τ is the pure lag time constant.

Further, the process of identifying the parameters in the SNCR denitration control system model by the IPSO algorithm in step 3 is shown in fig. 1, and specifically includes:

step 3.1: parameter initialization in IPSO algorithm, including population size D₁Number of iterations N₁Velocity v interval, inertia weight ω interval, learning factor C₁Learning factor C₂And the value ranges of an open loop gain coefficient K, an inertia time constant T, a system order n and a pure lag time constant tau in the SNCR denitration control system model;

step 3.2: calculating the fitness value of each particle in the initial population by taking the derivative of the mean square error function as a fitness function, and finding out the individual optimal value and the population optimal value in the initial population;

step 3.3: comparing the individual optimal value and the population optimal value in the initial population;

step 3.4: updating the position information and the speed information of the particles to obtain a new population;

step 3.5: calculating and comparing the individual optimal value and the group optimal value in the new population again;

step 3.6: when the maximum iteration number is met, optimizing is finished, K, T, n and tau after optimization are output, and otherwise, steps 3.2 to 3.6 are repeated.

Further, the population size D in step 3.1₁100, iteration number N ₁200, the speed interval v ∈ [ -1,1]The inertia weight interval ω ∈ [0.1,0.9 ]]Learning factor C₁＝C₂2.02, the open loop gain factor K e-5, 5]The inertia time constant T ∈ [1,400 ]]The system order n ∈ [1,4 ]]Pure lag time constant τ e [0,150 ∈ ]]；

In step 3.4, the speed information is iteratively updated according to the following formula:

V_ij(t+Δt)＝ωV_ij(t)+C₁R₁[X_bestij-X_ij(t)]+C₂R₂[X_bestgj-X_ij(t)]

in the formula X_bestijIs the optimum position of the ith row and jth column of particles, X_bestgjThe optimal position of the group is obtained; c₁And C₂Is a learning factor; t is the current time, and t plus delta t is the time after delta t at the time t; r₁And R₂Is a random factor, R₁,R₂∈[0,1]；X_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.. said., 100, j 1,2, a.. said., 4; ω is the inertial weight, and the expression is as follows:

wherein ω is_maxIs the maximum inertia weight, omega_minIs the minimum inertia weight, k₁For the current number of iterations, k_1maxIs the maximum number of iterations.

The optimal position of the particles is determined by:

in the formula X_bestiIs the optimal position of the particles; q_bestiAn optimal fitness value for the particle; t is the current time, and t plus delta t is the time after delta t at the time t; x_iThe position of the ith particle.

The position information is iteratively updated according to the following formula:

X_ij(t+Δt)＝X_ij(t)+V_ij(t+Δt)

wherein t is the current time, and t plus delta t is the time after delta t at the time t; x_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.

Further, the transfer function model of the urea solution flow to the nitrogen oxide concentration under the 170MW condition in step 3 is:

the model of the transfer function from the urea solution flow to the nitrogen oxide concentration under the 260MW working condition is as follows:

further, the adaptive genetic algorithm in step 4 is shown in fig. 2, and specifically includes the following steps:

step 4.1: initializing population, including K in PID controller_p、K_i、K_dValue ranges of three parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.2: coding the parameters in the step 4.1 by adopting a real number coding and linear interpolation mode;

step 4.3: selecting a fitness function Q (x) and calculating the fitness value of each individual;

step 4.4: selecting individuals meeting preset relative fitness value indexes by adopting a roulette method;

step 4.5: and (4) performing genetic operation on the individuals selected in the step 4.4, wherein the genetic operation formula is as follows:

wherein A, B is the individual selected by roulette method, A ', B' are the new individuals generated by genetic manipulation, α, β are constants, α, β ∈ [0,1 ].

Step 4.6: carrying out mutation operation on the individuals generated in the steps 4.4 and 4.5 by adopting a reverse method;

step 4.7: obtaining a new population after selection, heredity and mutation operations;

step 4.8: parameter decoding of new population, including K in PID controller_p、K_i、K_dThree parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.9: outputting the optimized K when the maximum iteration number is satisfied_p、K_i、K_dOtherwise, repeating steps 4.4 to 4.9.

Further, K in said step 4.1_p∈[-8,8]，K_i∈[1,10]，K_d∈[0,100]Population size D ₂50, iteration number N₂100, chromosome length L20, crossover probability P_c∈[0.4,0.9]The expression is as follows:

in the formula P_cmax，P_cminRespectively the maximum value and the minimum value of the cross probability; k is a radical of₂For the current number of iterations, k_2maxIs the maximum number of iterations.

Probability of variation P_mThe expression is as follows:

P_m＝0.2-[1:N₂]*0.04/N₂

the fitness function q (x) in step 4.3 has the following expression:

in the formula of omega₁、ω₂、ω₃Is constant, and ω₁+ω₂+ω₃1, e (t) is the deviation, t_sTo adjust the time, σ is the overshoot.

Further, as shown in fig. 3, the Smith estimation compensation control method in step 4 is to connect the Smith estimation compensator in parallel to the PID controller in the reverse direction on the basis of the original single-loop PID control loop, and use the SNCR denitration control system model as the model of the Smith estimation compensator.

The invention will be described inAdapting optimized K in genetic algorithms_p、K_i、K_dThe three parameters are used as the three parameters of the PID controller, and then are combined with a Smith estimation compensation control method, so that the output of the concentration of the nitrogen oxide in the control model of the SNCR denitration system is controlled.

Fig. 4 and 5 show the identification results under 170MW and 260MW operating conditions, the error indexes of the identification results are detailed in table 1, and the following mean square error MSE and root mean square error RMSE formulas are adopted when the error analysis is performed on the original data value of the nox concentration and the identification data value of the nox concentration output by the identification model:

wherein, y_iThe nitrogen oxide concentration original data value is obtained;

the identification data value is output by the identification model; m is the total number of samples.

TABLE 1 error index under identification of two typical conditions

The method comprises the steps of establishing a single-loop PID control and adaptive genetic algorithm Smith estimation compensation control model in a Simulink platform in MATLAB, and establishing NO_xThe concentration was set to a target value of 40mg/m³Adding external disturbance at 700s, with the size of 20mg/m³The control performance curves per unit step at 170MW and 260MW were obtained, see FIGS. 6 and 7.

As shown in tables 2 and 3, the simulation results were qualitatively analyzed in terms of rise time, adjustment time, overshoot, and steady-state error. The calculation result shows that under two typical working conditions, the overshoot under the AGA-Smith estimation compensation control is smaller than that under the single-loop PID control, which is quite beneficial to the injection quantity of the urea flow in the field, and when external disturbance is added, the overshoot generated by the AGA-Smith estimation compensation control is also small, and the regulation speed is high. The AGA-Smith predicted compensation control performance is generally superior to single loop PID control.

TABLE 2170 MW Performance index

Performance index	Single loop PID control	AGA-Smith predictive compensation control
			Rise time tr/s	54.85	129
Adjusting the time ts/s	358	612
			Overshoot σ/%)	51.123	10.52
Steady state error e ss	0	0

Control Performance index at 3260 MW

Performance index	Single loop PID control	AGA-Smith predictive compensation control
			Rise time tr/s	43.08	61.83
Adjusting the time ts/s	310	340
			Overshoot σ/%)	23.23	11.74
Steady state error e ss	0	0

Robustness verification

Although the IPSO algorithm is used for identifying the transfer function model of the SNCR denitration system urea flow under the typical working conditions of 170MW and 260MW, the field model parameters are not constant all the time and are influenced by the bed temperature, the limestone adding amount and other factors, the model parameters are bound to change, and the SNCR denitration system urea flow model under the condition of 170MW is used as verification under the condition that the system order is constant:

(1) t and tau are unchanged, K is changed, model is changedIs composed of

The control performance curve is shown in FIG. 8:

(2) t and K are unchanged, T is changed, and the model becomes

The control performance curve is shown in FIG. 9:

(3) k and τ are unchanged, T is changed, and the model becomes

The control performance curve is shown in FIG. 10:

when T and tau are changed and the control performance curve model parameters are disturbed and changed, the model adaptive capacity of single-loop PID control is lower than the adaptive capacity of AGA-Smith estimation compensation control. The introduction of the AGA-Smith estimation compensation control is effective in controlling the concentration of the nitrogen oxide, and provides a good technical reference for the actual concentration of the nitrogen oxide on site.

Claims (10)

1. A method for controlling the concentration of nitrogen oxides in a thermal power plant is characterized by comprising the following steps:

step 1: collecting urea solution flow x of unit under 170MW and 260MW working conditions from DCS system_iWith nitrogen oxide concentration y_iData;

step 2: the flow rate x of the urea solution_iWith nitrogen oxide concentration y_iCarrying out normalization processing, filtering processing and zero initial value processing on the data in sequence;

and step 3: identifying parameters in the SNCR denitration control system model by using an IPSO algorithm based on the data processed in the step 2 to obtain a transfer function model from urea solution flow to nitric oxide concentration under the working conditions of 170MW and 260 MW;

and 4, step 4: and (3) on the basis of single-loop PID control, a method of combining an adaptive genetic algorithm with a Smith prediction compensation control method is introduced to control the transfer function model obtained in the step (3).

2. The method for controlling nitrogen oxide concentration in a thermal power plant according to claim 1, wherein the normalization in step 2 adopts the following formula:

wherein x is_iIs the flow rate of urea solution, x_min、x_maxMinimum and maximum values of the urea solution flow, x, respectively_i1The value is obtained after the urea solution flow is normalized; y is_iIs the concentration of nitrogen oxides, y_min、y_maxMinimum and maximum nitrogen oxide concentration, y_i1The nitrogen oxide concentration is normalized; 1,2,3.. 1800;

the filtering processing in the step 2 adopts the following formula:

x_i2＝α₁x_i1+(1-α)x_i1′

wherein the smoothing coefficient alpha₁∈[0,1]；x_i1Is' x_i1Predicted value of (a), x_i2The value is obtained after filtering the flow of the urea solution; y is_i2The nitrogen oxide concentration is obtained by filtering; 1,2,3.. 1800;

the zero initial value processing in the step 2 adopts the following formula:

wherein x is_i3The flow rate of the urea solution is obtained after zero initial value processing; y is_i3The nitrogen oxide concentration is obtained after zero initial value processing; 1,2,3.. 1800;N＝5。

3. the method for controlling the concentration of nitrogen oxides in the thermal power plant according to claim 1, wherein the SNCR denitration control system model in the step 3 is as follows:

wherein K is a system open loop gain coefficient; t is₁.....T_nIs the inertia time constant; n is the system order; τ is the pure lag time constant.

4. The method for controlling the concentration of nitrogen oxides in the thermal power plant according to claim 1, wherein the specific process of identifying the parameters in the SNCR denitration control system model by the IPSO algorithm in the step 3 is as follows:

step 3.1: parameter initialization in IPSO algorithm, including population size D₁Number of iterations N₁Velocity v interval, inertia weight ω interval, learning factor C₁Learning factor C₂And the value ranges of an open loop gain coefficient K, an inertia time constant T, a system order n and a pure lag time constant tau in the SNCR denitration control system model;

step 3.2: calculating the fitness value of each particle in the initial population by taking the derivative of the mean square error function as a fitness function, and finding out the individual optimal value and the population optimal value in the initial population;

step 3.3: comparing the individual optimal value and the population optimal value in the initial population;

step 3.4: updating the position information and the speed information of the particles to obtain a new population;

step 3.5: calculating and comparing the individual optimal value and the group optimal value in the new population again;

step 3.6: when the maximum iteration number is met, optimizing is finished, K, T, n and tau after optimization are output, and otherwise, steps 3.2 to 3.6 are repeated.

5. The method for controlling nitrogen oxide concentration in thermal power plant according to claim 4, wherein the population size D in step 3.1₁100, iteration number N₁200, the speed interval v ∈ [ -1,1]The inertia weight interval ω ∈ [0.1,0.9 ]]Learning factor C₁＝C₂2.02, the open loop gain factor K e-5, 5]The inertia time constant T ∈ [1,400 ]]The system order n ∈ [1,4 ]]Pure lag time constant τ e [0,150 ∈ ]]。

6. The method for controlling the concentration of nitrogen oxides in a thermal power plant according to claim 4, wherein the speed information in step 3.4 is updated iteratively according to the following formula:

V_ij(t+Δt)＝ωV_ij(t)+C₁R₁[X_bestij-X_ij(t)]+C₂R₂[X_bestgj-X_ij(t)]

in the formula X_bestijIs the optimum position of the ith row and jth column of particles, X_bestgjThe optimal position of the group is obtained; c₁And C₂Is a learning factor; t is the current time, and t plus delta t is the time after delta t at the time t; r₁And R₂Is a random factor, R₁,R₂∈[0,1]；X_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.. said., 100, j 1,2, a.. said., 4; ω is the inertial weight, and the expression is as follows:

wherein ω is_maxIs the maximum inertia weight, omega_minIs the minimum inertia weight, k₁For the current number of iterations, k_1maxIs the maximum iteration number;

the optimal position of the particles is determined by:

in the formula X_bestiIs the optimal position of the particles; q_bestiAn optimal fitness value for the particle; t is the current time, and t plus delta t is the time after delta t at the time t; x_iIs the position of the ith particle;

the position information is iteratively updated according to the following formula:

X_ij(t+Δt)＝X_ij(t)+V_ij(t+Δt)

wherein t is the current time, and t plus delta t is the time after delta t at the time t; x_ijIs the position of the ith row and jth column of particles, V_ijThe speed of the ith row and the jth column of particles; 1,2,3, a.

7. The method for controlling nitrogen oxide concentration in a thermal power plant according to claim 1, wherein the transfer function model of urea solution flow to nitrogen oxide concentration under 170MW in step 3 is:

the model of the transfer function from the urea solution flow to the nitrogen oxide concentration under the 260MW working condition is as follows:

8. the method for controlling nitrogen oxide concentration in a thermal power plant according to claim 1, wherein the adaptive genetic algorithm in the step 4 specifically comprises the following steps:

step 4.1: initializing population, including K in PID controller_p、K_i、K_dValue ranges of three parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.2: coding the parameters in the step 4.1 by adopting a real number coding and linear interpolation mode;

step 4.3: selecting a fitness function Q (x) and calculating the fitness value of each individual;

step 4.4: selecting individuals meeting preset relative fitness value indexes by adopting a roulette method;

step 4.5: and (4) performing genetic operation on the individuals selected in the step 4.4, wherein the genetic operation formula is as follows:

a, B is selected by roulette method, A 'and B' are new individuals generated by genetic manipulation, alpha and beta are constants, and alpha, beta belongs to [0,1 ];

step 4.6: carrying out mutation operation on the individuals generated in the steps 4.4 and 4.5 by adopting a reverse method;

step 4.7: obtaining a new population after selection, heredity and mutation operations;

step 4.8: parameter decoding of new population, including K in PID controller_p、K_i、K_dThree parameters, population size D₂Number of iterations N₂Chromosome length L, crossover probability P_cAnd the mutation probability P_m；

Step 4.9: outputting the optimized K when the maximum iteration number is satisfied_p、K_i、K_dOtherwise, repeating steps 4.4 to 4.9.

9. The method for controlling nitrogen oxide concentration in thermal power plant according to claim 8, wherein K in step 4.1_p∈[-8,8]，K_i∈[1,10]，K_d∈[0,100]Population size D₂50, iteration number N₂100, chromosome length L20, crossover probability P_c∈[0.4,0.9]The expression is as follows:

in the formula P_cmax、P_cminRespectively the maximum value and the minimum value of the cross probability; k is a radical of₂For the current number of iterations, k_2maxIs the maximum iteration number;

probability of variation P_mThe expression is as follows:

P_m＝0.2-[1:N₂]*0.04/N₂

the fitness function q (x) in step 4.3 has the following expression:

in the formula of omega₁、ω₂、ω₃Is constant, and ω₁+ω₂+ω₃1, e (t) is the deviation, t_sTo adjust the time, σ is the overshoot.

10. The method for controlling the concentration of nitrogen oxides in a thermal power plant according to claim 1, wherein the Smith estimation compensation control method in the step 4 is that a Smith estimation compensator is reversely connected in parallel to a PID controller on the basis of an original single-loop PID control loop, and an SNCR denitration control system model is used as a model of the Smith estimation compensator.

Images