CN104190254A - Method for optimizing SCR (Selective Catalytic Reduction) ammonia spraying - Google Patents

Abstract

The invention relates to a method for optimizing SCR (Selective Catalytic Reduction) ammonia spraying. The method comprises the following steps: (1) predicting the SCR denitration efficiency of a machine unit according to a support vector machine; (2) calculating the total denitration cost according to the predicted denitration efficiency and the measured NOx production amount, boiler load, flue gas oxygen content, flue gas quantity and ammonia spraying amount; and (3) optimizing the ammonia spraying amount, thereby obtaining the ammonia spraying amount corresponding to the minimum total denitration cost. According to the method, the denitration efficiency of the machine unit is firstly predicted according to the support vector machine, then the total denitration cost is calculated according to the obtained denitration efficiency and the NOx generation amount, and finally, a model is optimized based on overall consideration of the denitration efficiency and the total denitration cost, the optimal ammonia spraying amount is acquired, and thus, the denitration cost is reduced.

Description

A kind of method of optimizing SCR spray ammonia

Technical field

The invention belongs to SCR (SCR) technical field of power plant's generating, especially a kind of method of optimizing SCR spray ammonia.

Background technology

The rapid growth of nitrogen oxide (NOx) discharge has aggravated the degradating trend of region acid rain, and result of study shows, the increase of nitrogen oxide emission make China's Acid Rain Pollution by sulfuric acid type to sulfuric acid and the compound transformation of nitric acid.Can estimate, coal steam-electric plant smoke denitration will become the emphasis of next step power plants efforts at environmental protection.

SCR (SCR) technology is the mainstream technology of current coal steam-electric plant smoke denitration, selective catalytic reduction denitrating technique technology, be utilize reducing agent under catalyst action selectively with flue gas in nitrogen oxide (NO _x, be mainly NO and NO ₂) there is chemical reaction, generate harmless nitrogen G&W, thereby remove NO in flue gas _xmethod.But SCR technological investment is large, and operation expense is high, adopting SCR technology to carry out denitrating flue gas will increase the cost of electricity power enterprise.So denitration economy is all very important for environmental protection and power plant.

Because denitration efficiency is subject to being permitted multifactorial impact, temporarily there is no concrete computing formula, except direct measurement, predict it is also a kind of feasible method by SVMs.

Current spray ammonia control system great majority only consider how to improve denitration efficiency, reduce the escaping of ammonia rate, nobody proposes directly taking denitration cost minimization as constraint, control the method for ammonia spraying amount, therefore, need a kind of SCR spray ammonia strategy taking denitration cost minimization as constraint, by optimizing, draw ammonia amount required under minimum denitration cost, thereby obtain best ammonia spraying amount, reduce NO _xdischarge, improves economy of power plant.

Summary of the invention

The object of the invention is in order to overcome the deficiencies in the prior art, a kind of method of the SCR of optimization spray ammonia is provided.

The present invention solves its technical problem and takes following technical scheme to realize:

1, a method of optimizing SCR spray ammonia, is characterized in that comprising that step is as follows:

(1) according to SVMs, the SCR denitration efficiency of prediction unit; Concrete Forecasting Methodology is:

1. taking boiler operatiopn data as basis, select the following variable that affects denitration efficiency: represent the generation load of boiler load, represent the burner hearth oxygen amount of excess air coefficient, ammonia spraying amount, NO _xgrowing amount is as input variable, and output variable is Denitration in Boiler efficiency;

2. the N group data of choosing different load section are used for training, are used for input parameter and the output parameter of modeling sample to be expressed as { x _i, y _i} ⁿ, x _ibe i group input parameter, y _ibe i group output parameter, N is sample size,

Adopt support vector machine method modeling, SVMs kernel function is selected RBF:

$K(x_i,x_j)=\mathrm{\φ}\left(x_i\right)\·\mathrm{\φ}\left(x_j\right)=\exp \left|\left(\frac{{\left|\right|x_i-x_j\left|\right|}^2}{{2\mathrm{\σ}}^2}\right)\right|$

Parameter σ is the width of RBF, and φ (x) is mapping function, and establishing required object function is f (x _i)=w φ (x _i)+b, f (x _i) be model predication value, w is weight coefficient vector, b is intercept.Introduce relaxation factor ξ _i ^*and ξ _iand permission error of fitting ε, ξ _i ^*>=0, ξ _i>=0, constraints is:

$\left\{\begin{array}{c}{y}_i-w\·\mathrm{\φ}\left(x_i\right)-b\≤\mathrm{\ϵ}+{\mathrm{\ξ}}_i\\ w\·\mathrm{\φ}\left(x_i\right)+b-y_i\≤\mathrm{\ϵ}+{{\mathrm{\ξ}}_i}^{*}\\ {{\mathrm{\ξ}}_i}^{*}\≥0\\ {\mathrm{\ξ}}_i\≥0,i=1,\·\·\·\·\·\·N\end{array}\right.$

Model is established as:

$\min R(w,\mathrm{\ξ},{\mathrm{\ξ}}^{*})=\frac{1}{2}w\·w+c\underset{i=1}{\overset{k}{\mathrm{\Σ}}}\mathrm{\ξ}+{\mathrm{\ξ}}^{*}$

Wherein constant c is penalty coefficient, c >=0; Introduce Lagrangian

$L(w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*},\mathrm{\α},{\mathrm{\α}}^{*},\mathrm{\γ},{\mathrm{\γ}}^{*})=\frac{1}{2}w\·w+C\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(\mathrm{\ξ}+{\mathrm{\ξ}}^{*})-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i[y_i-({\mathrm{\ξ}}_i+\mathrm{\ϵ}+f\left(x_i\right))]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{{\mathrm{\α}}_i}^{*}[{{\mathrm{\ξ}}_i}^{*}+\mathrm{\ϵ}+f\left(x_i\right)-y_i]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\γ}}_i{\mathrm{\ξ}}_i+{{\mathrm{\γ}}_i}^{*}{{\mathrm{\ξ}}_i}^{*})$

Wherein: α _i, α _i ^*, γ _i, γ _i ^*for Lagrange multiplier, be all greater than zero,

Ask function about w, b, ξ _i, ξ _i ^*minimum point, obtain:

$\left\{\begin{array}{c}\frac{\∂}{\∂w}L=0\⇒w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right)\\ \frac{\∂}{\∂b}L=0\⇒\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})=0\\ \frac{\∂}{{\∂\mathrm{\ξ}}_i}L=0\⇒C-{\mathrm{\α}}_i-{\mathrm{\γ}}_i=0\\ \frac{\∂}{{{\∂\mathrm{\ξ}}_i}^{*}}L=0\⇒C-{{\mathrm{\α}}_i}^{*}-{{\mathrm{\γ}}_i}^{*}=0\end{array}\right.$

Draw Lagrangian dual function:

$\stackrel{\‾}{\mathrm{\ω}}{(\mathrm{\α},{\mathrm{\α}}^{*})}_{w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*}}=-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})({\mathrm{\α}}_j-{{\mathrm{\α}}_j}^{*})K(x_i,x_j)-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})\mathrm{\ϵ}+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})y_i$

Now:

$w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right),f\left(x\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})K(x,x_i)+b$

According to KKT conditional theorem, at extreme point, place has:

α _i[ε+ξ _i-y _i+f(x _i)]＝0

α _i ^*[ε+ξ _i+y _i-f(x _i)]＝0 i＝1,……,N

Draw with this:

ξ _iγ _i＝0、ξ _i ^*γ _i ^*＝0 i＝1，……,N

Thereby draw b, obtain the relational model between input quantity and output quantity;

(2) according to denitration efficiency out of prediction and the NO of actual measurement _xgrowing amount, boiler load, flue gas oxygen content, exhaust gas volumn, ammonia spraying amount calculate denitration totle drilling cost; Circular is:

By the method for linear fit, set up the relational model of denitration device electric current and ammonia spraying amount;

Total current=k × ammonia spraying amount+b

Wherein coefficient k and b utilize institute's image data linear fit to obtain;

Calculate denitration totle drilling cost, comprising:

1. denitration power consumption cost: according to the denitration device electric current of above-mentioned foundation and the relational model of ammonia spraying amount, draw each current of electric summation by ammonia spraying amount, thereby calculate power consumption, calculate power consumption in conjunction with electricity price and go out cost;

Consumed power=1.732 × equipment voltage × equipment total current × power factor

Power consumption expense=power consumption × inner electricity price

Simultaneous three formulas obtain:

Power consumption expense=1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price

2. ammoniacal liquor consuming cost:

Ammoniacal liquor consuming cost=ammonia spraying amount × ammoniacal liquor price

3. pollution charge: according to the NO of the denitration efficiency doping and actual measurement _xgrowing amount, calculates NO _xdischarge capacity, in conjunction with blowdown and the expenses standard of national regulation, expenses standard that can the unit's of obtaining blowdown flow rate, uses thereby calculate charges for disposing pollutants;

NO _xdischarge capacity=NO _xgrowing amount × (1-denitration efficiency)

Charges for disposing pollutants=NO _xdischarge capacity × expenses standard=NO _xgrowing amount × (1-denitration efficiency) × expenses standard

Wherein, expenses standard specifies according to environmental administration, in conjunction with NO _xdischarge capacity draws;

4. electricity price compensation: according to the denitration efficiency doping and actual measurement NO _xgrowing amount, calculates NO _xdischarge capacity, in conjunction with national electricity price ethical policy, can obtain denitration effect and whether can reach compensation requirement, thereby draws whether enjoy electricity price compensation;

Totle drilling cost=ammoniacal liquor consuming cost+denitration power consumption cost+pollution charge-electricity price compensation=ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xthe compensation of growing amount × (1-denitration efficiency) × expenses standard-electricity price

Add up to cost:

Wherein, the model prediction gained that denitration efficiency is set up according to SVMs, is about load, ammonia spraying amount, Oxygen Amount in Flue Gas and NO _xthe function of growing amount;

NOx growing amount × [fine of SCR denitration efficiency × ammoniacal liquor price+(1-SCR denitration efficiency) × discharge NOx] of cost=Σ coal i ratio × coal i of NOx;

Wherein, i is that the N kind that unit burning is used is mixed the i kind in coal;

(3) ammonia spraying amount is optimized, finally obtains hour corresponding ammonia spraying amount of denitration totle drilling cost;

With:

Min (totle drilling cost)=min (ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xgrowing amount × (1-denitration efficiency) × expenses standard-electricity price compensation)

For target, adopt without the lower steepest descent method of constraint and optimize ammonia spraying amount, finally obtain the ammonia spraying amount of hour correspondence of denitration totle drilling cost.

Advantage of the present invention and good effect are:

First the present invention utilizes SVMs, and the denitration efficiency of prediction unit, then according to gained denitration efficiency and NO _xgrowing amount, calculates denitration totle drilling cost, finally considers denitration efficiency and each denitration cost, and model is optimized, and draws best ammonia spraying amount, and then reduces out-of-stock cost.

Detailed description of the invention

Below the invention process is further described, following examples are descriptive, are not determinate, can not limit protection scope of the present invention with this.

A method of optimizing SCR spray ammonia, the step of the method comprises:

(1) primarily utilize SVMs, the SCR denitration efficiency of prediction unit; Concrete Forecasting Methodology is:

1. taking boiler operatiopn data as basis, select the following variable that affects denitration efficiency: represent the generation load of boiler load, represent the burner hearth oxygen amount of excess air coefficient, ammonia spraying amount, NO _xgrowing amount is as input variable, and output variable is Denitration in Boiler efficiency;

2. the N group data of choosing different load section are used for training, are used for input parameter and the output parameter of modeling sample to be expressed as { x _i, y _i} ⁿ, x _ibe i group input parameter, y _ibe i group output parameter, N is sample size,

Adopt support vector machine method modeling, SVMs kernel function is selected RBF:

$K(x_i,x_j)=\mathrm{\φ}\left(x_i\right)\·\mathrm{\φ}\left(x_j\right)=\exp \left|\left(\frac{{\left|\right|x_i-x_j\left|\right|}^2}{{2\mathrm{\σ}}^2}\right)\right|$

Parameter σ is the width of RBF, and φ (x) is mapping function, and establishing required object function is f (x _i)=w φ (x _i)+b, f (x _i) be model predication value, w is weight coefficient vector, b is intercept.Introduce relaxation factor ξ _i ^*and ξ _iand permission error of fitting ε, ξ _i ^*>=0, ξ _i>=0, constraints is:

$\left\{\begin{array}{c}{y}_i-w\·\mathrm{\φ}\left(x_i\right)-b\≤\mathrm{\ϵ}+{\mathrm{\ξ}}_i\\ w\·\mathrm{\φ}\left(x_i\right)+b-y_i\≤\mathrm{\ϵ}+{{\mathrm{\ξ}}_i}^{*}\\ {{\mathrm{\ξ}}_i}^{*}\≥0\\ {\mathrm{\ξ}}_i\≥0,i=1,\·\·\·\·\·\·N\end{array}\right.$

Model is established as:

$\min R(w,\mathrm{\ξ},{\mathrm{\ξ}}^{*})=\frac{1}{2}w\·w+c\underset{i=1}{\overset{k}{\mathrm{\Σ}}}\mathrm{\ξ}+{\mathrm{\ξ}}^{*}$

Wherein constant c is penalty coefficient, c >=0;

Introduce Lagrangian

$L(w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*},\mathrm{\α},{\mathrm{\α}}^{*},\mathrm{\γ},{\mathrm{\γ}}^{*})=\frac{1}{2}w\·w+C\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(\mathrm{\ξ}+{\mathrm{\ξ}}^{*})-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i[y_i-({\mathrm{\ξ}}_i+\mathrm{\ϵ}+f\left(x_i\right))]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{{\mathrm{\α}}_i}^{*}[{{\mathrm{\ξ}}_i}^{*}+\mathrm{\ϵ}+f\left(x_i\right)-y_i]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\γ}}_i{\mathrm{\ξ}}_i+{{\mathrm{\γ}}_i}^{*}{{\mathrm{\ξ}}_i}^{*})$

Wherein: α _i, α _i ^*, γ _i, γ _i ^*for Lagrange multiplier, be all greater than zero,

Ask function about w, b, ξ _i, ξ _i ^*minimum point, obtain:

$\left\{\begin{array}{c}\frac{\∂}{\∂w}L=0\⇒w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right)\\ \frac{\∂}{\∂b}L=0\⇒\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})=0\\ \frac{\∂}{{\∂\mathrm{\ξ}}_i}L=0\⇒C-{\mathrm{\α}}_i-{\mathrm{\γ}}_i=0\\ \frac{\∂}{{{\∂\mathrm{\ξ}}_i}^{*}}L=0\⇒C-{{\mathrm{\α}}_i}^{*}-{{\mathrm{\γ}}_i}^{*}=0\end{array}\right.$

Draw Lagrangian dual function:

$\stackrel{\‾}{\mathrm{\ω}}{(\mathrm{\α},{\mathrm{\α}}^{*})}_{w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*}}=-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})({\mathrm{\α}}_j-{{\mathrm{\α}}_j}^{*})K(x_i,x_j)-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})\mathrm{\ϵ}+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})y_i$

Now:

$w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right),f\left(x\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})K(x,x_i)+b$

According to KKT conditional theorem, at extreme point, place has:

α _i[ε+ξ _i-y _i+f(x _i)]＝0

α _i ^*[ε+ξ _i+y _i-f(x _i)]＝0 i＝1,……,N

Draw with this:

ξ _iγ _i＝0、ξ _i ^*γ _i ^*＝0 i＝1,……,N

Thereby draw b, obtain the relational model between input quantity and output quantity;

(2) according to denitration efficiency out of prediction and the NO of actual measurement _xgrowing amount, boiler load, flue gas oxygen content, exhaust gas volumn, ammonia spraying amount calculate denitration totle drilling cost; Circular is:

By the method for linear fit, set up the relational model of denitration device electric current and ammonia spraying amount;

Total current=k × ammonia spraying amount+b

Wherein coefficient k and b utilize institute's image data linear fit to obtain;

Calculate denitration totle drilling cost, comprising:

A. denitration power consumption cost: according to the denitration device electric current of above-mentioned foundation and the relational model of ammonia spraying amount, draw each current of electric summation by ammonia spraying amount, thereby calculate power consumption, calculate power consumption in conjunction with electricity price and go out cost;

Consumed power=1.732 × equipment voltage × equipment total current × power factor

Power consumption expense=power consumption × inner electricity price

Simultaneous three formulas obtain:

Power consumption expense=1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price

Totle drilling cost=ammoniacal liquor consuming cost+denitration power consumption cost+pollution charge-electricity price compensation=ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xthe compensation of growing amount × (1-denitration efficiency) × expenses standard-electricity price

B. ammoniacal liquor consuming cost:

Ammoniacal liquor consuming cost=ammonia spraying amount × ammoniacal liquor price

C. pollution charge: according to the NO of the denitration efficiency doping and actual measurement _xgrowing amount, calculates NO _xdischarge capacity, in conjunction with blowdown and the expenses standard of national regulation, expenses standard that can the unit's of obtaining blowdown flow rate, uses thereby calculate charges for disposing pollutants;

NO _xdischarge capacity=NO _xgrowing amount × (1-denitration efficiency)

Charges for disposing pollutants=NO _xdischarge capacity × expenses standard=NO _xgrowing amount × (1-denitration efficiency) × expenses standard

Wherein, expenses standard specifies according to environmental administration, in conjunction with NO _xdischarge capacity draws;

D. electricity price compensation: according to the denitration efficiency doping and actual measurement NO _xgrowing amount, calculates NO _xdischarge capacity, in conjunction with national electricity price ethical policy, can obtain denitration effect and whether can reach compensation requirement, thereby draws whether enjoy electricity price compensation;

Add up to cost:

Wherein, the model prediction gained that denitration efficiency is set up according to SVMs, is about load, ammonia spraying amount, Oxygen Amount in Flue Gas and NO _xthe function of growing amount;

Be exemplified below: NOx growing amount × [fine of SCR denitration efficiency × ammoniacal liquor price+(1-SCR denitration efficiency) × discharge NOx] of cost=Σ coal i ratio × coal i of NOx;

Wherein, i is that the N kind that unit burning is used is mixed the i kind in coal;

Ammonia spraying amount is optimized, finally obtains hour corresponding ammonia spraying amount of denitration totle drilling cost;

With:

Min (totle drilling cost)=min (ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xgrowing amount × (1-denitration efficiency) × expenses standard-electricity price compensation)

For target, adopt without the lower steepest descent method of constraint and optimize ammonia spraying amount, finally obtain hour corresponding ammonia spraying amount of denitration totle drilling cost, thereby realize the object of raising economy of power plant.

Example

Taking certain power plant as example, first predict denitration efficiency: select 20 groups of training samples to train, obtain denitration efficiency and load, ammonia spraying amount, flue gas oxygen content and NO _xrelational model between growing amount.

After model has been set up, selecting different operating modes is example, prediction denitration efficiency, and result is as follows.

Totle drilling cost=ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xthe compensation of growing amount × (1-denitration efficiency) × expenses standard-electricity price

Under the different operating modes of table 1, denitration efficiency predicts the outcome

Regard ammonia spraying amount as variable to be optimized, other parameters are regarded definite value as, draw the function of cost about ammonia spraying amount according to totle drilling cost computing formula:

In this example, according to actual conditions, the price of ammoniacal liquor is 2680 yuan/t, and denitration device voltage is 6000V, and power factor gets 0.83,0.48 yuan/kwh of inner electricity price; The expenses standard of the discharge NOx calculating is 5.95 yuan/kg;

Adopt steepest descent method to be optimized ammonia spraying amount, show that best ammonia spraying amount after optimization is as following table,

Table 2 is optimized rear ammonia spraying amount

Claims (4)

1. a method of optimizing SCR spray ammonia, is characterized in that comprising that step is as follows:

(1) according to SVMs, the SCR denitration efficiency of prediction unit; Concrete Forecasting Methodology is:

1. taking boiler operatiopn data as basis, select the following variable that affects denitration efficiency: represent the generation load of boiler load, represent the burner hearth oxygen amount of excess air coefficient, ammonia spraying amount, NO _xgrowing amount is as input variable, and output variable is Denitration in Boiler efficiency;

2. the N group data of choosing different load section are used for training, are used for input parameter and the output parameter of modeling sample to be expressed as { x _i, y _i} ⁿ, x _ibe i group input parameter, y _ibe i group output parameter, N is sample size,

Adopt support vector machine method modeling, SVMs kernel function is selected RBF:

$K(x_i,x_j)=\mathrm{\φ}\left(x_i\right)\·\mathrm{\φ}\left(x_j\right)=\exp \left|\left(\frac{{\left|\right|x_i-x_j\left|\right|}^2}{{2\mathrm{\σ}}^2}\right)\right|$

Parameter σ is the width of RBF, and φ (x) is mapping function, and establishing required object function is f (x _i)=w φ (x _i)+b, f (x _i) be model predication value, w is weight coefficient vector, b is intercept.Introduce relaxation factor ξ _i ^*and ξ _iand permission error of fitting ε, ξ _i ^*>=0, ξ _i>=0, constraints is:

$\left\{\begin{array}{c}{y}_i-w\·\mathrm{\φ}\left(x_i\right)-b\≤\mathrm{\ϵ}+{\mathrm{\ξ}}_i\\ w\·\mathrm{\φ}\left(x_i\right)+b-y_i\≤\mathrm{\ϵ}+{{\mathrm{\ξ}}_i}^{*}\\ {{\mathrm{\ξ}}_i}^{*}\≥0\\ {\mathrm{\ξ}}_i\≥0,i=1,\·\·\·\·\·\·N\end{array}\right.$

Model is established as:

$\min R(w,\mathrm{\ξ},{\mathrm{\ξ}}^{*})=\frac{1}{2}w\·w+c\underset{i=1}{\overset{k}{\mathrm{\Σ}}}\mathrm{\ξ}+{\mathrm{\ξ}}^{*}$

Wherein constant c is penalty coefficient, c >=0;

Introduce Lagrangian

$L(w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*},\mathrm{\α},{\mathrm{\α}}^{*},\mathrm{\γ},{\mathrm{\γ}}^{*})=\frac{1}{2}w\·w+C\underset{i=1}{\overset{N}{\mathrm{\Σ}}}(\mathrm{\ξ}+{\mathrm{\ξ}}^{*})-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_i[y_i-({\mathrm{\ξ}}_i+\mathrm{\ϵ}+f\left(x_i\right))]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{{\mathrm{\α}}_i}^{*}[{{\mathrm{\ξ}}_i}^{*}+\mathrm{\ϵ}+f\left(x_i\right)-y_i]-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\γ}}_i{\mathrm{\ξ}}_i+{{\mathrm{\γ}}_i}^{*}{{\mathrm{\ξ}}_i}^{*})$

Wherein: α _i, α _i ^*, γ _i, γ _i ^*for Lagrange multiplier, be all greater than zero,

Ask function about w, b, ξ _i, ξ _i ^*minimum point, obtain:

$\left\{\begin{array}{c}\frac{\∂}{\∂w}L=0\⇒w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right)\\ \frac{\∂}{\∂b}L=0\⇒\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})=0\\ \frac{\∂}{{\∂\mathrm{\ξ}}_i}L=0\⇒C-{\mathrm{\α}}_i-{\mathrm{\γ}}_i=0\\ \frac{\∂}{{{\∂\mathrm{\ξ}}_i}^{*}}L=0\⇒C-{{\mathrm{\α}}_i}^{*}-{{\mathrm{\γ}}_i}^{*}=0\end{array}\right.$

Draw Lagrangian dual function:

$\stackrel{\‾}{\mathrm{\ω}}{(\mathrm{\α},{\mathrm{\α}}^{*})}_{w,b,\mathrm{\ξ},{\mathrm{\ξ}}^{*}}=-\frac{1}{2}\underset{i,j=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})({\mathrm{\α}}_j-{{\mathrm{\α}}_j}^{*})K(x_i,x_j)-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})\mathrm{\ϵ}+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i+{{\mathrm{\α}}_i}^{*})y_i$

Now:

$w=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})\mathrm{\φ}\left(x_i\right),f\left(x\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}({\mathrm{\α}}_i-{{\mathrm{\α}}_i}^{*})K(x,x_i)+b$

According to KKT conditional theorem, at extreme point, place has:

α _i[ε+ξ _i-y _i+f(x _i)]＝0

α _i ^*[ε+ξ _i+y _i-f(x _i)]＝0 i＝1,……,N

Draw with this:

ξ _iγ _i＝0、ξ _i ^*γ _i ^*＝0 i＝1,……,N

Thereby draw b, obtain the relational model between input quantity and output quantity;

(2) according to denitration efficiency out of prediction and the NO of actual measurement _xgrowing amount, boiler load, flue gas oxygen content, exhaust gas volumn, ammonia spraying amount calculate denitration totle drilling cost; Circular is:

By the method for linear fit, set up the relational model of denitration device electric current and ammonia spraying amount;

Total current=k × ammonia spraying amount+b

Wherein coefficient k and b utilize institute's image data linear fit to obtain;

Calculate denitration totle drilling cost, comprising:

1. denitration power consumption cost: according to the denitration device electric current of above-mentioned foundation and the relational model of ammonia spraying amount, draw each current of electric summation by ammonia spraying amount, thereby calculate power consumption, calculate power consumption in conjunction with electricity price and go out cost;

Consumed power=1.732 × equipment voltage × equipment total current × power factor

Power consumption expense=power consumption × inner electricity price

Simultaneous three formulas obtain:

Power consumption expense=1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price

2. ammoniacal liquor consuming cost:

Totle drilling cost=ammoniacal liquor consuming cost+denitration power consumption cost+pollution charge-electricity price compensation=ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xthe compensation of growing amount × (1-denitration efficiency) × expenses standard-electricity price

Ammoniacal liquor consuming cost=ammonia spraying amount × ammoniacal liquor price;

3. pollution charge: according to the NO of the denitration efficiency doping and actual measurement _xgrowing amount, calculates NO _xdischarge capacity, in conjunction with blowdown and expenses standard, the expenses standard of the unit's of obtaining blowdown flow rate, uses thereby calculate charges for disposing pollutants;

NO _xdischarge capacity=NO _xgrowing amount × (1-denitration efficiency)

Charges for disposing pollutants=NO _xdischarge capacity × expenses standard=NO _xgrowing amount × (1-denitration efficiency) × expenses standard;

4. the cost of NOx: according to the denitration efficiency doping and actual measurement NO _xgrowing amount, calculates NO _xdischarge capacity cost;

Add up to NOx growing amount × [fine of SCR denitration efficiency × ammoniacal liquor price+(1-SCR denitration efficiency) × discharge NOx] of cost=Σ coal i ratio × coal i of cost: NOx;

Wherein, i is that the N kind that unit burning is used is mixed the i kind in coal;

(3) ammonia spraying amount is optimized, in conjunction with electricity price compensation, finally obtains hour corresponding ammonia spraying amount of denitration totle drilling cost;

With:

Min (totle drilling cost)=min (ammonia spraying amount × ammoniacal liquor price+1.732 × equipment voltage × (k × ammonia spraying amount+b) × power factor × inner electricity price+NO _xgrowing amount × (1-denitration efficiency) × expenses standard-electricity price compensation)

For target, adopt without the lower steepest descent method of constraint and optimize ammonia spraying amount, finally obtain the ammonia spraying amount of hour correspondence of denitration totle drilling cost.

2. the method for optimization SCR spray ammonia according to claim 1, is characterized in that: in the 3. step of described step (2), expenses standard is according to environmental administration's regulation, in conjunction with NO _xdischarge capacity draws.

3. the method for optimization SCR spray ammonia according to claim 1, is characterized in that: specifically country's NOx blowdown and expenses standard at that time of fine of the 4. middle discharge NOx of described step (2).

4. the method for optimization SCR spray ammonia according to claim 1, is characterized in that: the electricity price in described step (3) is compensated for as country's electricity price ethical policy at that time.