CN105809297A - Thermal power plant environment economic dispatching method based on multi-target differential evolution algorithm - Google Patents

Abstract

The invention discloses a thermal power plant environment economic dispatching method based on a multi-target differential evolution algorithm.The method comprises the following steps that a thermal power plant economic dispatching model with the lowest electricity generation cost and smallest pollutant discharge quantity as targets and with generator capacity and power balance as constraint conditions is built; the multi-target differential evolution algorithm is utilized for carrying out optimization solving on the model, an optimal Pareto solution set is obtained, the multi-target differential evolution algorithm adopts difference mutation operators for searching, mutation operators are selected based on the accumulation performance and using frequency of the operators of the latest several times of variation, and solution set convergence and distribution uniformity are ensured by means of non-dominated ranking, domination frequency and hypervolume contribution and the like; finally, a decision is made through the fuzzy set theory, and a compromise solution is selected from the Pareto solution set to be used as a final regulation scheme.The thermal power plant environment economic dispatching method has the advantages that precision is high, Pareto leading edge solution set distribution is uniform and convergence speed is high, and engineering realization is easy.

Description

A kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm

Technical field

The present invention relates to electric power system dispatching technical field, particularly to the thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm.

Background technology

Economic Dispatch is to meet on Operation of Electric Systems constraints basis, solving the scheduling scheme making cost of electricity-generating network minimal.For thermoelectric generator, it will give off dusty gas or the greenhouse gases such as substantial amounts of oxysulfide, nitrogen oxides and carbon dioxide in power generation process.If it is considered that pollutant discharge amount in power generation process, originally the economic load dispatching of single goal is changed into the scheduling of multiobject environmental economy.Owing to the goal of cost and pollutant emission target collide with each other, therefore add the difficulty formulating scheduling scheme.

The method solving environmental economy scheduling totally can be divided into mathematical programming approach and the big class of intelligent optimization method two.Mathematic programming methods mainly adopts leash law and the means such as weight coefficient and method that Multi-Objective Scheduling is converted into single goal scheduling problem, by repeatedly solving acquisition Pareto disaggregation, compatibility is poor, computational efficiency is relatively low for the extension of this kind of method, and due to environmental economy scheduling be a nonlinearity optimization problem, it is thus achieved that Pareto disaggregation be frequently not global optimum.Intelligent optimization method single solves and can obtain Pareto disaggregation, have that realization is simple, optimize that efficiency is high, be suitable for solving the features such as nonlinear problem, it is widely used in solving of environmental economy scheduling problem, such as genetic algorithm, particle cluster algorithm, bacterial foraging algorithm, random search etc..Differential evolution algorithm is a kind of Intelligent Optimization Technique with simplicity, concurrency and strong robustness, and it carries out optimizing by special mutation operation in search volume, is to solve for the effective means of environmental economy scheduling.It addition, no matter be mathematical programming approach or intelligent optimization method, after obtaining Pareto disaggregation, how to carry out reasonably decision-making and select certain compromise solution as final embodiment to concentrate from solution, be also one of key realizing thermal power plant's environmental economy scheduling.

At present, thermal power plant's environmental economy scheduling was carried out some researchs by educational circles, and retrieval section achievement, " adopting the environmental economy power generation dispatching of multi-target improvement differential evolution algorithm " such as " the power system environment economic load dispatching based on multi-objective Evolutionary Algorithm " of Zhu Yongsheng et al. and " adopting based on the electric power environmental economic load dispatching of multi-objective Evolutionary Algorithm decomposed ", Hu Bin et al. etc., but it is accurate not still to there is Difference Calculation in said method, it is impossible to improve and solve the problem that environmental economy sends out the feasibility in a scheduling and effectiveness.

Summary of the invention

For the problems referred to above, it is an object of the invention to provide a kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm, overcome existing method to be difficult to obtain the deficiency of the Pareto disaggregation with truncation strategy and distributing homogeneity when solving this problem, and carry out Rational Decision and realize the selection of final embodiment.

The present invention solves its technical problem, the technical scheme adopted is:

A kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm, comprises the following steps:

Step 1: set up thermal power plant's environmental economy scheduling mathematic model

Thermal power plant's economic load dispatching mathematical model includes object function and constraints, and wherein object function is described by following formula:

$\begin{array}{cc}min& \[\underset{k=1}{\overset{K}{\Σ}}{F}_k\left(P_k\right),\underset{k=1}{\overset{K}{\Σ}}{E}_k\left(P_k\right)\]\end{array},$

Wherein, K is electromotor number, P in thermal power generation unit_kFor kth platform electromotor active power of output, F_k(P_k) it is P for kth platform electromotor active power of output_kShi Suoxu generates electricity expense, E_k(P_k) be i-th electromotor active power of output it is P_kTime pollutant discharge amount, k=1,2 ..., K, generate electricity expense F_k(P_k) calculation expression be:

F_k(P_k)=a_k+b_kP_k+c_kP_k ²,

Wherein, a_k、b_kAnd c_kFor the cost coefficient of kth platform electromotor, k=1,2 ..., K, description is the expenses such as fuel needed for electrical power generators, labour force and plant maintenance, is known parameters for concrete electromotor；

Pollutant discharge amount E_k(P_k) calculation expression be:

E_k(P_k)=α_k+β_kP_k+γ_kP_k ²+ξ_kexp(λ_kP_k),

Wherein, α_k、β_k、γ_k、ξ_kAnd λ_kRespectively kth platform electromotor pollutant emission characterisitic parameter, k=1,2 ..., K, is known parameters for concrete thermoelectric generator；

The constraints of thermal power plant's environmental economy scheduling includes generator capacity constraint and power-balance constraint；Generator capacity constraint is described by following formula:

$P_k^{min}\≤P_k\≤P_k^{max},$

Wherein,WithThe respectively minima of kth platform electromotor active power of output and maximum, k=1,2 ..., K, for the setting value of selected electromotor；Power-balance is constrained to the total output of generating set should be equal to via net loss and bearing power sum, and expression formula is:

$\underset{k=1}{\overset{K}{\Σ}}{P}_k=P_D+P_L,$

Wherein,For the active power of output that K platform electromotor in thermal power generation unit is total, P_DFor system load power, P_LFor via net loss power.P_DFor known parameters, P_LCan be calculated by following B coefficient method:

$P_L=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}{P}_i{B}_{ij}{P}_j+\underset{i=1}{\overset{K}{\Σ}}{B}_{0i}{P}_i+B_{00},$

Wherein, B_ij、B_0iAnd B₀₀For via net loss coefficient, i=1,2 ..., K, j=1,2 ..., K, is known parameters for concrete power system；

Step 2: all kinds of parameters required in model in obtaining step 1, including: the bearing power P that in electrical network, system is total_D, computing network loss power parameter B_ij、B_0iAnd B₀₀, the minimum and maximum active power of output of each electromotor in thermal power generation unitWithGenerating set cost parameters a_k、b_kAnd c_i, generating set pollutant discharge amount parameter alpha_k、β_k、γ_k、ξ_kAnd λ_k, i=1,2 ..., K, j=1,2 ..., K, k=1,2 ..., K；

Step 3: adopt multiple target differential evolution algorithm that the mathematical model of step 1 is optimized and solve, it is thus achieved that Pareto disaggregationN is the number solving and concentrating scheduling scheme, solves and concentrates P^g,iRepresent i-th kind of scheduling scheme, i=1,2 ..., N；

Step 4: on the Pareto disaggregation basis that step 3 obtains, utilizes Fuzzy Set Theory method to calculate the satisfaction of each scheduling scheme, and its calculation expression is:

${\mathrm{\μ}}_n=\frac{{\mathrm{\μ}}_n^F+{\mathrm{\μ}}_n^E}{\underset{n=1}{\overset{N}{\Σ}}({\mathrm{\μ}}_n^F+{\mathrm{\μ}}_n^E)},$

Wherein, μ_nRepresent Pareto solution and concentrate the total satisfaction of the n-th scheduling scheme,Represent the satisfaction of the goal of cost in the n-th scheduling scheme,Represent the satisfaction of pollutant discharge amount in the n-th scheduling scheme, n=1,2 ..., N.WithCan be calculated by following formula respectively:

${\mathrm{\&mu;}}_n^F=\left\{\begin{array}{c}1,F_n\&le;F_{min}\\ \frac{F_{max}-F_n}{F_{\max }-F_{\min }},F_{min}<F_n<F_{max}\\ 0,F_n\&GreaterEqual;F_{\max }\end{array}\right.,$

${\mathrm{\&mu;}}_n^E=\left\{\begin{array}{c}1,E_n\&le;E_{\min }\\ \frac{E_{max}-E_n}{E_{\max }-E_{min}},E_{min}<E_n<E_{max}\\ 0,E_n\&GreaterEqual;E_{max}\end{array}\right.,$

Wherein, F_nAnd E_nRepresent the n-th scheduling scheme generating expense and pollutant discharge amount, F respectively_minAnd F_maxRepresent minima and maximum, the E of the expense that generates electricity in all N number of scheduling schemes respectively_minAnd E_maxRepresent minima and the maximum of pollutant discharge amount in all N number of scheduling schemes respectively；

Step 5: the scheduling scheme satisfaction that step 4 is obtainedNumerical value compare sequence, therefrom select the maximum scheduling scheme of satisfaction numerical value, as final scheduling scheme, and selected scheduling scheme numerical value corresponded in the Pareto forward position scattergram drawn according to this method.

Further, the acquired all kinds of model parameters of described step 2 depend on concrete thermal power plant's generating set, for some thermal power generation unit, if being left out exponential term ξ therein when calculating regulation goal functional value_kexp(λ_kP_k), then can make ξ_kAnd λ_kIt is 0, j=1,2 ..., K, k=1,2 ..., K, this step utilizes the parameter obtained that concrete thermal power plant is carried out Model instantiation.

Further, described step 3 comprises the following steps:

Step 3.1, initiation parameter NP, W, C, G, α, β；Iteration count initial value g=1 and q is set_l=0, n_l=0, l=1,2,3；Initialize the disaggregation being made up of NP initial solution, also referred to as candidate solution, be designated as P^g={ P^g,1,P^g,2,…,P^g,NP, each initial solutionRepresent a kind of scheduling scheme,Represent the active power of output of kth platform electromotor in i-th scheduling scheme, time initial?In randomly choose, i=1,2 ..., NP, k=1,2 ..., K；

Step 3.2, candidate solution are likely to be unsatisfactory for power-balance constraint, and the solution for not satisfying the constraint condition adopts heuristic to be modified, and revise step and are: for candidate solutionI=1,2 ..., NP, first with B Y-factor method Y computing network lossAnd calculate this candidate solution constraint amount of running counter to according to following formula:

$P_v^{g,i}=P_D+P_L^{g,i}-\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k^{g,i},$

IfBe not zero, then randomly choose from K electromotor an electromotor k ∈ 1,2 ..., K}, output it powerCurrency increasesThen right according to generator capacity constraintCarry out bound constrained process, be namely set to spanIn boundary value or random value；Recalculate via net loss on this basisIf revised candidate solution still runs counter to power-balance constraint, then repeat above-mentioned makeover process, until it reaches certain reparation number of times or P_cvTerminate when value is less than a certain less threshold value；

Step 3.3, first calculating candidate scheme P^g,iIn the expense of each electromotorAnd pollutant discharge amountK=1,2 ..., K；Then time scheme P is calculated^g,iTotal costAnd gross contamination emissionI=1,2 ..., NP；

Step 3.4, for disaggregation P^gIn i-th candidate solution P^g,i, from the 3 of differential evolution algorithm kinds of mutation operators, { 1,2,3} to P for selection opertor r ∈^g,iCarry out mutation operation, produce variable V^g,i, update the access times n of r operator in nearest W mutation operation simultaneously_r, 3 kinds of differential evolution operators are respectively depicted as:

V^g,i=P^g,r1+α·(P^g,r1-P^g,r1),

V^g,i=P^g,r1+α·(P^g,r2-P^g,r3)+α·(P^g,r4-P^g,r5),

V^g,i=P^g,i+rand(0,1)·(P^g,r1-P^g,i)+α·(P^g,r2-P^g,r3),

In formula, α is the parameter performing corresponding computing, P^g,r1、P^g,r2、P^g,r3、P^g,r4And P^g,r5For participating in the candidate solution of mutation operator in differential evolution operator, they are from disaggregation P^gIn randomly choose but different, rand (0,1) is the random number between 0 and 1；Wherein, the system of selection of operator r is: if existed in 3 operators from the operator being not used by, then randomly choosing one from the operator being not used by, if 3 operators all used, then selecting by following formula:

$r=\underset{l=1,2,3}{\mathrm{argmax}}(q_l+C\sqrt{\frac{2log\left({\mathrm{\&Sigma;}}_{l=1}^3{n}_l\right)}{n_l}}),$

Wherein, n_lFor the access times of l operator, q in nearest W mutation operation_lFor the accumulation performance of operator l in nearest W mutation operation, C is constant.If P^g,iAdopt the 1st kind or the 2nd kind of mutation operator carries out computing, then proceed as follows further: first produce random integers k_rand∈ 1,2 ..., K}, then for P^g,iIn each variablePerform the intersection operation described by following formula, produce new candidate solution U^g,i:

Wherein,WithRepresent variable V respectively^g,iAnd U^g,iIn kth variable, β for intersect operation desired parameters；If P^g,iThe 3rd kind of operator is adopted to carry out computing, then by new candidate solution U^g,iValue is V^g,i；

If step 3.5 candidate solution U^g,iIt is unsatisfactory for generator capacity constraint and power-balance constraint, then adopts the method described by step 3.2 to be modified, calculate revised candidate solution U^g,iCorresponding cost value and pollutant discharge amount；Then by U^g,iJoin disaggregation P^gIn, now P^gInterior candidate solution number is NP+1；

Step 3.6, according to new disaggregation P^gIn the performance of each candidate solution, delete 1 worst solution, make population number maintain NP；Delete step is: carry out non-dominated ranking according to the cost value of all candidate solutions and pollutant discharge amount, by P^gIt is divided into F₁,F₂,...,F_LIn forward position；If forward position number of plies L is more than 1 and F_LIn only have 1 solution, then directly delete this solution；If forward position number of plies F_LMore than 1 and F_LIn have multiple solution, then delete F_LThe middle solution maximum by other candidate solution domination number of times；If forward position number of plies L is equal to 1, then delete each candidate solution and there is the solution that hypervolume contribution value is minimum, candidate solution hypervolume contribution amount is defined as the area in all candidate solutions and the enclosed region of reference point and deducts and reject the area in residue candidate solution and the enclosed region of reference point after this candidate solution, and wherein reference point each desired value value is the maximum of this desired value in all candidate solutions；

Step 3.7, calculation procedure 3.4 are used the operator effect γ after the r mutation operator_r, and update the accumulation performance q of each operator in nearest W mutation operation_l, l=1,2,3, γ_rEmploying following formula calculates:

γ_r=1/m₁·1/(m₂+1)+m₃,

Wherein, m₁For the candidate solution U using the r mutation operator to produce in step 3.4^g,iForward position sequence number residing during non-bad sequence in step 3.6, m₂For the candidate solution U using the r mutation operator to produce in step 3.4^g,iThe number of times arranged by other candidate solution；m₃For candidate solution U^g,iHypervolume contribution amount；Accumulation performance q_lComputational methods be: for nearest all W mutation operations, W effect value is ranked up from small to large, remembers that the w effect is ordered as R_w, the delay ranking value DR of w effect is then calculated according to following formula_w, it may be assumed that

DR_w=(W-R_w+1)·D^W-w,

The DR calculated after using according to W operator_wValue calculates the AUC area under curve of each operator further, calculates each operator accumulation performance q according to following formula on this basis_l, l=1,2,3:

$q_l=\frac{{\mathrm{AUC}}_l}{{\mathrm{\&Sigma;}}_{l=1}^3{\mathrm{AUC}}_l},$

Wherein, AUC_lIt it is the area in the enclosed region of ROC curve of l operator；

Step 3.8, repetition step 3.4～step 3.7, until P^gIn each candidate solution carried out mutation operation；

Step 3.9, iterations enumerator g value add 1, repeat step 3.4～step 3.8, until iterations enumerator g value reaches maximum iteration time G；Then to disaggregation P^gCarry out non-bad sequence, P^gIn all non-domination solution constitute Pareto disaggregationN is disaggregationThe number of middle solution.

Compared with prior art, the method have the advantages that

One, the present invention realizes simply, fast operation, strong robustness, the heuristics manner being easily achieved is adopted to realize power-balance constraint in search procedure, and consider in power-balance and to utilize via net loss, calculated by B Y-factor method Y and try to achieve, and simple bound constrained method realizes generator capacity constraint in search procedure；

Two, the present invention is by adopting multiple differential evolution operator collaboratively searchings, bad sequence and the domination measure such as number of times statistics to guarantee the convergence of Pareto disaggregation in multiple target differential evolution algorithm, and wherein differential evolution operator selects to be realize according to the access times of each operator in nearest several times mutation operation and accumulation performance thereof；And adopt hypervolume contribution amount strategy to guarantee the uniformity that Pareto disaggregation is distributed, provide optimal candidate scheme for follow-up decision；

Three, have on truncation strategy and distributing homogeneity Pareto disaggregation basis in acquisition, utilize Fuzzy Set Theory to provide, for policymaker, the dispatching method that satisfaction is the highest, it is achieved the effect of thermal power plant's generating expense and pollutant discharge amount comprehensively optimum.

In a word, the present invention adopts simple constraint processing method to realize thermal power generation unit capacity-constrained and power-balance constraint；Adopt many differential evolution operators collaboratively searching and guarantee convergence and the distributing homogeneity of Pareto disaggregation in search procedure based on strategies such as hypervolume contribution amount, overcoming the deficiency utilizing crowding method to be difficult to accurately keep disaggregation distributing homogeneity in general multi-objective Algorithm；Adopt Fuzzy Set Theory to concentrate the scheme selecting most satisfaction as embodiment from Pareto solution, there is stronger practicality.

Accompanying drawing explanation

Fig. 1 is the FB(flow block) of the present invention；

Fig. 2 is the Pareto forward position scattergram that obtains of the embodiment of the present invention and the compromise solution of selection；

Fig. 3 is the Pareto forward position scattergram that obtains of GDE3 method and the compromise solution of selection；

Fig. 4 is the present invention and GDE3 method optimal cost desired value convergence curve；

Fig. 5 is the present invention and GDE3 method optimum pollutant discharge amount desired value convergence curve.

Detailed description of the invention

Below in conjunction with the embodiment in accompanying drawing, the present invention is described in further detail, but is not intended that any limitation of the invention.

A kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm, initially sets up thermal power plant's environmental economy scheduling mathematic model, and obtains the model parameter of concrete thermal power plant generating set；Then utilize multiple target differential evolution algorithm that this model is solved, it is thus achieved that Pareto forward position disaggregation；Calculate Pareto forward position on this basis and solve the satisfaction concentrating each candidate's scheduling scheme, and select the scheduling scheme of Maximum Satisfaction as final scheduling scheme.Flow chart of the present invention is as it is shown in figure 1, specifically include following steps:

Step 1, set up the mathematical model of thermal power plant environmental economy scheduling

Thermal power plant's economic load dispatching mathematical model includes object function and constraints, and wherein object function is described by following formula:

$\begin{array}{cc}min& \&lsqb;\underset{k=1}{\overset{K}{\&Sigma;}}{F}_k\left(P_k\right),\underset{k=1}{\overset{K}{\&Sigma;}}{E}_k\left(P_k\right)\&rsqb;\end{array},$

Wherein, K is electromotor number, P in thermal power generation unit_kFor kth platform electromotor active power of output, F_k(P_k) it is P for kth platform electromotor active power of output_kShi Suoxu generates electricity expense, E_k(P_k) it is P for kth platform electromotor active power of output_kTime pollutant discharge amount, k=1,2 ..., K, generate electricity expense F_k(P_k) calculation expression be:

$F_k\left(P_k\right)=a_k+b_k{P}_k+c_k{P}_k^2,$

Wherein, a_k、b_kAnd c_kFor the cost coefficient of kth platform electromotor, k=1,2 ..., K, description is the expenses such as fuel needed for electrical power generators, labour force and plant maintenance, is known parameters for concrete electromotor；Pollutant discharge amount E_k(P_k) calculation expression be:

$E_k\left(P_k\right)={\mathrm{\&alpha;}}_k+{\mathrm{\&beta;}}_k{P}_k+{\mathrm{\&gamma;}}_k{P}_k^2+{\mathrm{\&xi;}}_k\exp \left({\mathrm{\&lambda;}}_k{P}_k\right),$

Wherein, α_k、β_k、γ_k、ξ_kAnd λ_kRespectively kth platform electromotor pollutant emission characterisitic parameter, k=1,2 ..., K, is known parameters for concrete thermoelectric generator.

The constraints of thermal power plant's environmental economy scheduling includes generator capacity constraint and power-balance constraint；Generator capacity constraint is described by following formula:

$P_k^{min}\&le;P_k\&le;P_k^{max},$

Wherein,WithThe respectively minima of kth platform electromotor active power of output and maximum, k=1,2 ..., K, is known parameters for concrete electromotor；Power-balance is constrained to the total output of generating set should be equal to via net loss and bearing power sum, and expression formula is:

$\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k=P_D+P_L,$

Wherein,For the active power of output that K platform electromotor in thermal power generation unit is total, P_DFor system load power, P_LFor via net loss power；P_DFor known parameters, P_LCan be calculated by following B coefficient method:

$P_L=\underset{i=1}{\overset{K}{\&Sigma;}}\underset{j=1}{\overset{K}{\&Sigma;}}{P}_i{B}_{ij}{P}_j+\underset{i=1}{\overset{K}{\&Sigma;}}{B}_{0i}{P}_i+B_{00},$

Wherein, B_ij、B_0iAnd B₀₀For via net loss coefficient, i=1,2 ..., K, j=1,2 ..., K, is known parameters for concrete power system；

For some thermal power plant's environmental economy scheduling problem, retrain P discounting for network loss_L, then power-balance constraint only need to meetCalculate network loss without using B Y-factor method Y, namely now can make P_LIt is 0.This mathematical model is the basis that subsequent step carries out Multiobjective Scheduling；

Step 2, all kinds of parameters obtained in model, including the bearing power P that system in electrical network is total_D, computing network loss power parameter B_ij、B_0iAnd B₀₀, the minimum and maximum active power of output of each electromotor in thermal power generation unitWithGenerating set cost parameters a_k、b_kAnd c_i, generating set pollutant discharge amount parameter alpha_k、β_k、γ_k、ξ_kAnd λ_k, i=1,2 ..., K, j=1,2 ..., K, k=1,2 ..., K.

The acquired all kinds of model parameters of this step depend on concrete thermal power plant's generating set, for some thermal power generation unit, if being left out exponential term ξ therein when calculating regulation goal functional value_kexp(λ_kP_k), then can make ξ_kAnd λ_kIt is 0, j=1,2 ..., K, k=1,2 ..., K；This step utilizes the parameter obtained that concrete thermal power plant is carried out Model instantiation；

Step 3: adopt multiple target differential evolution algorithm that mathematical model is optimized and solve, it is thus achieved that Pareto disaggregationN is the number solving and concentrating scheduling scheme, solves and concentrates P^g,iRepresent i-th kind of scheduling scheme, i=1,2 ..., N；

This step obtains the Pareto disaggregation of concrete thermal power plant economic environment scheduling, this disaggregation represents some excellent solutions, but owing to being absent from directly good and bad relation between these solutions, namely two solutions this solution concentrated, it is assumed to be A and B, solve and an A object function wherein is better than solving B, but solution B will be inferior on another object function；Owing to Pareto solution concentrates the number solved to be limited, simultaneously for obtaining more excellent solution to guarantee the performance finally selecting compromise solution, the convergence that should have of Pareto disaggregation obtained and distributing homogeneity, namely Pareto disaggregation will the true Pareto forward position of approximation problem as far as possible, all solutions to be uniformly distributed as much as possible in Pareto forward position simultaneously；

Step 4: on the Pareto disaggregation basis that step 3 obtains, utilizes Fuzzy Set Theory method to calculate the satisfaction of each scheduling scheme, and its calculation expression is:

${\mathrm{\&mu;}}_n=\frac{{\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E}{\underset{n=1}{\overset{N}{\&Sigma;}}({\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E)},$

Wherein, μ_nRepresent Pareto solution and concentrate the total satisfaction of the n-th scheduling scheme,Represent the satisfaction of the goal of cost in the n-th scheduling scheme,Represent the satisfaction of pollutant discharge amount in the n-th scheduling scheme, n=1,2 ..., N.WithCan be calculated by following formula respectively:

${\mathrm{\&mu;}}_n^F=\left\{\begin{array}{c}1,F_n\&le;F_{min}\\ \frac{F_{max}-F_n}{F_{max}-F_{\min }},F_{min}<F_n<F_{max}\\ 0,F_n\&GreaterEqual;F_{\max }\end{array}\right.,$

${\mathrm{\&mu;}}_n^E=\left\{\begin{array}{c}1,E_n\&le;E_{\min }\\ \frac{E_{max}-E_n}{E_{\max }-E_{min}},E_{min}<E_n<E_{max}\\ 0,E_n\&GreaterEqual;E_{max}\end{array}\right.,$

Wherein, F_nAnd E_nRepresent the n-th scheduling scheme generating expense and pollutant discharge amount, F respectively_minAnd F_maxRepresent minima and maximum, the E of the expense that generates electricity in all N number of scheduling schemes respectively_minAnd E_maxRepresent minima and the maximum of pollutant discharge amount in all N number of scheduling schemes respectively；

In described step 3, the Pareto solution of acquisition is concentrated, and cannot be made directly and compare between each solution, and this step adopts satisfaction to describe the overall good and bad degree of each solution quantitatively, is the foundation of next step acquisition compromise solution；

Step 5: obtain on the basis of all scheduling scheme satisfactions in step 4, selects the scheduling scheme of Maximum Satisfaction as final scheduling scheme；

This step is satisfied with angle value based on the Pareto solution each solution of concentration, concentrates the solution selecting have maximum satisfaction as final scheduling scheme from Pareto solution, and what the program represented is an optimal compromise of the goal of cost and pollutant discharge amount.

Further, described step 3 adopts multiple target differential evolution algorithm mathematical model is optimized the detailed process solved is:

Step 3.1, initiation parameter NP, W, C, G, α, β；Iteration count initial value g=1 and q is set_l=0, n_l=0, l=1,2,3；Initialize the disaggregation being made up of NP initial solution, also referred to as candidate solution, be designated as P^g={ P^g,1,P^g,2,…,P^g,NP, each initial solutionRepresent a kind of scheduling scheme,Represent the active power of output of kth platform electromotor in i-th scheduling scheme, time initial?In randomly choose, i=1,2 ..., NP, k=1,2 ..., K；

Step 3.2, candidate solution are likely to be unsatisfactory for power-balance constraint, and the solution for not satisfying the constraint condition adopts heuristic to be modified, and revise step and are: for candidate solutionI=1,2 ..., NP, first with B Y-factor method Y computing network lossAnd calculate this candidate solution constraint amount of running counter to according to following formula:

$P_v^{g,i}=P_D+P_L^{g,i}-\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k^{g,i},$

IfBe not zero, then randomly choose from K electromotor an electromotor k ∈ 1,2 ..., K}, output it powerCurrency increasesThen right according to generator capacity constraintCarry out bound constrained process, be namely set to spanIn boundary value or random value；Recalculate via net loss on this basisIf revised candidate solution still runs counter to power-balance constraint, then repeat above-mentioned makeover process, until it reaches certain reparation number of times or P_cvTerminate when value is less than a certain less threshold value；

Step 3.3, first calculating candidate scheme P^g,iIn the expense of each electromotorAnd pollutant discharge amountK=1,2 ..., K；Then candidate scheme P is calculated^g,iTotal costAnd gross contamination emissionI=1,2 ..., NP；

Step 3.4, for disaggregation P^gIn i-th candidate solution P^g,i, from the 3 of differential evolution algorithm kinds of mutation operators, { 1,2,3} to P for selection opertor r ∈^g,iCarry out mutation operation, produce variable V^g,i, update the access times n of r operator in nearest W mutation operation simultaneously_r；3 kinds of differential evolution operators are respectively depicted as:

V^g,i=P^g,r1+α·(P^g,r1-P^g,r1),

V^g,i=P^g,r1+α·(P^g,r2-P^g,r3)+α·(P^g,r4-P^g,r5),

V^g,i=P^g,i+rand(0,1)·(P^g,r1-P^g,i)+α·(P^g,r2-P^g,r3),

In formula, α is the parameter performing corresponding computing, P^g,r1、P^g,r2、P^g,r3、P^g,r4And P^g,r5For participating in the candidate solution of mutation operator in differential evolution operator, they are from disaggregation P^gIn randomly choose but different, rand (0,1) is the random number between 0 and 1；Wherein, the system of selection of operator r is: if existed in 3 operators from the operator being not used by, then randomly choosing one from the operator being not used by, if 3 operators all used, then selecting by following formula:

$r=\underset{l=1,2,3}{\mathrm{argmax}}(q_l+C\sqrt{\frac{2log\left({\mathrm{\&Sigma;}}_{l=1}^3{n}_l\right)}{n_l}}),$

Wherein, n_lFor the access times of l operator, q in nearest W mutation operation_lFor the accumulation performance of operator l in nearest W mutation operation, C is constant.If P^g,iAdopt the 1st kind or the 2nd kind of mutation operator carries out computing, then proceed as follows further: first produce random integers k_rand∈ 1,2 ..., K}, then for P^g,iIn each variablePerform the intersection operation described by following formula, produce new candidate solution U^g,i:

Wherein,WithRepresent variable V respectively^g,iAnd U^g,iIn kth variable, β for intersect operation desired parameters；If P^g,iThe 3rd kind of operator is adopted to carry out computing, then by new candidate solution U^g,iValue is V^g,i；

If step 3.5 candidate solution U^g,iIt is unsatisfactory for generator capacity constraint and power-balance constraint, then adopts the method described by step 3.2 to be modified, calculate revised candidate solution U^g,iCorresponding cost value and pollutant discharge amount；Then by U^g,iJoin disaggregation P^gIn, now P^gInterior candidate solution number is NP+1；

Step 3.6, according to new disaggregation P^gIn the performance of each candidate solution, delete 1 worst solution, make population number maintain NP.Delete step is: carry out non-dominated ranking according to the cost value of all candidate solutions and pollutant discharge amount, by P^gIt is divided into F₁,F₂,...,F_LIn forward position；If forward position number of plies L is more than 1 and F_LIn only have 1 solution, then directly delete this solution；If forward position number of plies F_LMore than 1 and F_LIn have multiple solution, then delete F_LThe middle solution maximum by other candidate solution domination number of times；If forward position number of plies L is equal to 1, then delete each candidate solution and there is the solution that hypervolume contribution value is minimum, candidate solution hypervolume contribution amount is defined as the area in all candidate solutions and the enclosed region of reference point and deducts and reject the area in residue candidate solution and the enclosed region of reference point after this candidate solution, and wherein reference point each desired value value is the maximum of this desired value in all candidate solutions；

Step 3.7, calculation procedure 3.4 are used the operator effect γ after the r mutation operator_r, and update the accumulation performance q of each operator in nearest W mutation operation_l, l=1,2,3.γ_rEmploying following formula calculates:

γ_r=1/m₁·1/(m₂+1)+m₃,

Wherein, m₁For the candidate solution U using the r mutation operator to produce in step 3.4^g,iForward position sequence number residing during non-bad sequence in step 3.6, m₂For the candidate solution U using the r mutation operator to produce in step 3.4^g,iThe number of times arranged by other candidate solution；m₃For candidate solution U^g,iHypervolume contribution amount；Accumulation performance q_lComputational methods be: for nearest all W mutation operations, W effect value is ranked up from small to large, remembers that the w effect is ordered as R_w, the delay ranking value DR of w effect is then calculated according to following formula_w, it may be assumed that

DR_w=(W-R_w+1)·D^W-w,

The DR calculated after using according to W operator_wValue also utilizes area under curve method to draw the AUC curve of each operator, calculates each operator accumulation performance q according to following formula on this basis_l, l=1,2,3:

$q_l=\frac{{\mathrm{AUC}}_l}{{\mathrm{\&Sigma;}}_{l=1}^3{\mathrm{AUC}}_l},$

Wherein, AUC_lIt it is the area in the enclosed region of AUC curve of l operator；

Step 3.8, repetition step 3.4-step 3.7, until P^gIn each candidate solution carried out mutation operation；

Step 3.9, iterations enumerator g value add 1, repeat step 3.4-step 3.8, until iterations enumerator g value reaches maximum iteration time G；Then to disaggregation P^gCarry out non-bad sequence, P^gIn all non-domination solution constitute Pareto disaggregationN is disaggregationThe number of middle solution.

Embodiment

As it is shown in figure 1, apply method provided by the present invention, select IEEE30 node standard power systems as experimental subject, carry out environmental economy scheduling experiment, specifically comprise the following steps that

1) setting up the mathematical model of thermal power plant's environmental economy scheduling, for IEEE30 node system, mathematical model can whole description be:

$\begin{array}{cc}min& \&lsqb;\underset{k=1}{\overset{K}{\&Sigma;}}{F}_k\left(P_k\right),\underset{k=1}{\overset{K}{\&Sigma;}}{E}_k\left(P_k\right)\&rsqb;\end{array},$

F_k(P_k)=a_k+b_kP_k+c_kP_k ²,

E_k(P_k)=α_k+β_kP_k+γ_kP_k ²+ξ_kexp(λ_kP_k),

Generator capacity constraint and power-balance constraint should be met simultaneously, it may be assumed that

$P_k^{min}\&le;P_k\&le;P_k^{max}$

$\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k=P_D+P_L$

Wherein:

In above-mentioned model, K is thermoelectric generator number, P in generating set_kFor kth platform thermoelectric generator active power of output, F_k(P_k) it is P for kth platform thermoelectric generator active power of output_kShi Suoxu generates electricity expense, a_k、b_kAnd c_iIt is the cost coefficient of i-th thermoelectric generator, E_k(P_k) be i-th thermoelectric generator active power of output it is P_kTime pollutant discharge amount, α_k、β_k、γ_k、ξ_kAnd λ_kRespectively kth platform thermoelectric generator pollutant emission characterisitic parameter,WithThe respectively minima of kth platform electromotor active power of output and maximum, P_DFor system load power, P_LFor via net loss power, B_ij、B_0iAnd B₀₀For via net loss coefficient, i=1,2 ..., K, j=1,2 ..., K, k=1,2 ..., K；

2) all kinds of parameters in model, electromotor number K=6, system total load P in IEEE30 node modular system are obtained_D=283.4MW, Model Parameter and via net loss coefficient are respectively as shown in Table 1 and Table 2；

Table 1IEEE30 node modular system generating set data

Table 2IEEE30 node modular system via net loss coefficient

3) initiation parameter NP=50, W=20, C=3, G=200, α=0.5, β=0.9；G=1, q are set_l=0, n_l=0, l=1,2,3；Initialize disaggregation P^g={ P^g,1,P^g,2,…,P^g,50, each candidate solutionRepresent a kind of scheduling scheme,?In randomly choose, i=1,2 ..., 50, k=1,2 ..., 6；

4) candidate solution is likely to be unsatisfactory for power-balance constraint formula, and the solution for not satisfying the constraint condition adopts heuristic to be modified, and revises step and is: for candidate solutionI=1,2 ..., 50, first with loss factor B_ij、B_0iAnd B₀₀Computing network lossAnd calculate this candidate solution constraint amount of running counter to according to following formula:

$P_v^{g,i}=P_D+P_L^{g,i}-\underset{k=1}{\overset{6}{\&Sigma;}}{P}_k^{gj},$

IfBe not zero, then randomly choose from 6 electromotors an electromotor k ∈ 1,2 ..., 6}, output it powerCurrency increasesThen formula pair is retrained according to generator capacityCarry out bound constrained process, be namely set to spanIn boundary value or random value；Recalculate via net loss on this basisIf revised candidate solution still runs counter to power-balance constraint, then repeat above-mentioned makeover process, until it reaches certain reparation number of times or P_cvTerminate when value is less than a certain less threshold value；

5) each solution P that candidate solution is concentrated^g,i, i=1,2 ..., 50, calculate the expense of each electromotorAnd pollutant discharge amountK=1,2 ..., 6；Then scheme P is calculated^g,iTotal costAnd gross contamination emission

6) for disaggregation P^gIn i-th candidate solution P^g,i, from following 3 kinds of differential evolution algorithm mutation operators, { 1,2,3} to P for selection opertor r ∈^g,iCarry out computing, produce variable V^g,i, and update the access times n of r operator in nearest 20 mutation operations_r。

V^g,i=P^g,r1+α·(P^g,r1-P^g,r1),

V^g,i=P^g,r1+α·(P^g,r2-P^g,r3)+α·(P^g,r4-P^g,r5),

V^g,i=P^g,i+rand(0,1)·(P^g,r1-P^g,i)+α·(P^g,r2-P^g,r3),

In formula, α=0.5, P^g,r1、P^g,r2、P^g,r3、P^g,r4And P^g,r5For participating in the candidate solution of computing in differential evolution operator, they are by disaggregation P^gIn randomly choose but different, rand (0,1) is the random number between 0 and 1.Operator r system of selection is: if there is, in 3 operators, the operator being not used by, then randomly choosing one from the operator being not used by, if 3 operators all used, then selecting by following formula:

$r=\underset{l=1,2,3}{\mathrm{argmax}}(q_l+C\sqrt{\frac{2log\left({\mathrm{\&Sigma;}}_{l=1}^3{n}_l\right)}{n_l}}),$

Wherein, C=3, n_lFor the access times of l operator, q in nearest 20 mutation operations_lFor the accumulation performance of operator l in nearest 20 all mutation operations.If P^g,iAdopt the 1st kind or the 2nd kind of operator makes a variation, then first produce random integers k_rand∈ 1,2 ..., 6}, then for P^g,iIn each variablePerform to intersect and operate, produce new candidate solution U^g,i, it may be assumed that

Wherein, β=0.9,WithRepresent variable V respectively^g,iAnd U^g,iIn kth variable；If P^g,iThe 3rd kind of operator is adopted to make a variation, then by new candidate solution U^g,iValue is V^g,i；

7) if candidate solution U^g,iBe unsatisfactory for constraint formula, then according to step 4) described by method be modified, calculate revised candidate solution U^g,iCorresponding cost value and pollutant discharge amount；Then by U^g,iJoin disaggregation P^gIn, i.e. P^gInterior candidate solution number is 51；

8) according to new disaggregation P^gIn the performance of each candidate solution, delete 1 worst solution, make population number maintain 50.Delete step is: carry out non-bad sequence according to the cost value of all candidate solutions and pollutant discharge amount, by P^gIt is divided into F₁,F₂,...,F_LIn forward position；If forward position number of plies L is more than 1 and F_LIn only have 1 solution, then directly delete this solution；If forward position number of plies F_LMore than 1 and F_LIn have multiple solution, then delete F_LThe middle solution maximum by other candidate solution domination number of times；If forward position number of plies L is equal to 1, then delete each candidate solution and there is the solution that hypervolume contribution value is minimum, candidate solution hypervolume contribution amount is defined as the area in all candidate solutions and the enclosed region of reference point and deducts and reject the area in residue candidate solution and the enclosed region of reference point after this candidate solution, and wherein reference point each desired value value is the maximum of this desired value in all candidate solutions；

9) calculation procedure 6) in use the operator effect γ after the r mutation operator_r, and update the accumulation performance of each operator, i.e. q in nearest 20 mutation operations_l, l=1,2,3.Effect value γ_rEmploying following formula calculates:

γ_r=1/m₁·1/(m₂+1)+m₃,

Wherein, m₁For step 6) use the candidate solution U that the r mutation operator produce^g,iIn step 8) in non-bad sequence time residing forward position sequence number, m₂For step 6) the middle candidate solution U using the r mutation operator to produce^g,iThe number of times arranged by other candidate solution；m₃For candidate solution U^g,iHypervolume contribution margin；Accumulation performance q_lComputational methods be: for nearest 20 all of mutation operations, 20 effect value are ranked up from small to large, remember that the w effect is ordered as R_w, the delay ranking value DR of w effect is then calculated according to following formula_w, it may be assumed that

DR_w=(20-R_w+1)·D^20-w,

The DR calculated after using according to 20 operators_wValue and area under curve method draw the AUC curve of each operator, calculate each operator accumulation performance q according to following formula on this basis_l, l=1,2,3:

$q_l=\frac{{\mathrm{AUC}}_l}{{\mathrm{\&Sigma;}}_{l=1}^3{\mathrm{AUC}}_l},$

Wherein, AUC_lIt it is the area in the enclosed region of AUC curve of l operator；

10) step 6 is repeated)-step 9), until P^gIn each candidate solution carried out mutation operation；

11) iterations enumerator g value adds 1, repeats step 6)-step 10), until g=200；Then to disaggregation P^gCarry out non-bad sequence, P^gIn all non-domination solution constitute Pareto disaggregationN is disaggregationThe number of middle solution.

12) Fuzzy Set Theory method is utilized to calculateIn the satisfaction μ of each candidate's scheduling scheme_n, for the n-th scheme, μ_nComputational methods are:

${\mathrm{\&mu;}}_n=\frac{{\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E}{\underset{n=1}{\overset{N}{\&Sigma;}}({\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E)},$

${\mathrm{\&mu;}}_n^F=\{\begin{array}{c}1,F_n\&le;F_{min}\\ \frac{F_{max}-F_n}{F_{max}-F_{\min }}\\ 0,F_n\&GreaterEqual;F_{\max }\end{array},F_{min}<F_n<F_{max},$

${\mathrm{\&mu;}}_n^E=\left\{\begin{array}{c}1,E_n\&le;E_{\min }\\ \frac{E_{max}-E_n}{E_{\max }-E_{min}},E_{min}<E_n<E_{max}\\ 0,E_n\&GreaterEqual;E_{max}\end{array}\right.,$

Wherein, F_nAnd E_nRepresent the n-th scheduling scheme generating expense and pollutant discharge amount, F respectively_minAnd F_maxRepresent minima and maximum, the E of the expense that generates electricity in all scheduling schemes respectively_minAnd E_maxRepresent minima and the maximum of pollutant discharge amount in all scheduling schemes respectively.

13) from step 12) calculate all scheduling schemes in μ_nIn, select the scheduling scheme of Maximum Satisfaction as final scheduling scheme.

The present embodiment is calculated machine emulation, and compares with well-known multiple target differential evolution algorithm GDE3.Table 3, table 4 and table 5 sets forth the inventive method and GDE3 method optimal cost scheduling result, optimum pollutant emission scheduling result and compromise scheduling result.Fig. 2 and Fig. 3 sets forth the compromise solution of the inventive method Pareto forward position distribution with the acquisition of GDE3 method and selection, and Fig. 4 and Fig. 5 sets forth the inventive method and the scheduling of GDE3 method optimal cost and optimum discharge regulation goal value convergence curve.Utilize in the Pareto forward position scattergram as shown in Figure 2 obtained in the method for the present invention, find out the compromise solution point position of correspondence.

Dispatching for optimal cost, table 3 result shows that scheduling scheme expense of the present invention is 606.0130 $/h, is better than GDE3 method scheduling scheme expense 606.0323 $/h；For optimum discharge scheduling, table 4 result shows that scheduling scheme pollutant discharge amount of the present invention is 0.1942t/h, is slightly better than GDE3 method scheduling scheme pollutant discharge amount 0.1943t/h；For compromise solution, table 3 result shows that scheduling scheme expense of the present invention is 614.5114 $/h, and pollutant discharge amount is 0.2016t/h, is better than GDE3 method scheduling scheme.Meanwhile, Fig. 2 and Fig. 3 shows, the Pareto forward position that the inventive method obtains is more uniform compared with the Pareto forward position distribution that GDE3 method obtains；The convergence curve of Fig. 4 and Fig. 5 also shows that the inventive method convergence is also superior to GDE3 method.The above results fully shows that the present invention has precision height in thermal power plant's environmental economy scheduling problem, Pareto forward position disaggregation is evenly distributed and the characteristic of fast convergence rate.

Table 3 the inventive method and GDE3 method optimal cost scheduling result

Table 4 the inventive method and GDE3 method optimum discharge scheduling result

Table 5 the inventive method scheduling result compromise with GDE3 method

The above is only the preferred embodiment of the present invention, it is noted that the invention is not limited in aforesaid way, under the premise without departing from the principles of the invention, moreover it is possible to improve further, and these improvement also should be regarded as protection scope of the present invention.

Claims (3)

1. the thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm, it is characterised in that comprise the following steps:

Step 1: set up thermal power plant's environmental economy scheduling mathematic model

Thermal power plant's economic load dispatching mathematical model includes object function and constraints, and wherein object function is described by following formula:

$\min \&lsqb;\underset{k=1}{\overset{K}{\&Sigma;}}{F}_k\left(P_k\right),\underset{k=1}{\overset{K}{\&Sigma;}}{E}_k\left(P_k\right)\&rsqb;,$

Wherein, K is electromotor number, P in thermal power generation unit_kFor kth platform electromotor active power of output, F_k(P_k) it is P for kth platform electromotor active power of output_kShi Suoxu generates electricity expense, E_k(P_k) be i-th electromotor active power of output it is P_kTime pollutant discharge amount, k=1,2 ..., K, generate electricity expense F_k(P_k) calculation expression be:

$F_k\left(P_k\right)=a_k+b_k{P}_k+c_k{P}_k^2,$

Wherein, a_k、b_kAnd c_kFor the cost coefficient of kth platform electromotor, k=1,2 ..., K, description is the expenses such as fuel needed for electrical power generators, labour force and plant maintenance, is known parameters for concrete electromotor；

Pollutant discharge amount E_k(P_k) calculation expression be:

$E_k\left(P_k\right)={\mathrm{\&alpha;}}_k+{\mathrm{\&beta;}}_k{P}_k+{\mathrm{\&gamma;}}_k{P}_k^2+{\mathrm{\&xi;}}_k\exp \left({\mathrm{\&lambda;}}_k{P}_k\right),$

Wherein, α_k、β_k、γ_k、ξ_kAnd λ_kRespectively kth platform electromotor pollutant emission characterisitic parameter, k=1,2 ..., K, is known parameters for concrete thermoelectric generator；

The constraints of thermal power plant's environmental economy scheduling includes generator capacity constraint and power-balance constraint；Generator capacity constraint is described by following formula:

$P_k^{min}\&le;P_k\&le;P_k^{max},$

Wherein,WithThe respectively minima of kth platform electromotor active power of output and maximum, k=1,2 ..., K, for the setting value of selected electromotor；Power-balance is constrained to the total output of generating set should be equal to via net loss and bearing power sum, and expression formula is:

$\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k=P_D+P_L,$

Wherein,For the active power of output that K platform electromotor in thermal power generation unit is total, P_DFor system load power, P_LFor via net loss power；P_DFor known parameters, P_LCan be calculated by following B coefficient method:

$P_L=\underset{i=1}{\overset{K}{\&Sigma;}}\underset{j=1}{\overset{K}{\&Sigma;}}{P}_i{B}_{ij}{P}_j+\underset{i=1}{\overset{K}{\&Sigma;}}{B}_{0i}{P}_i+B_{00},$

Wherein, B_ij、B_0iAnd B₀₀For via net loss coefficient, i=1,2 ..., K, j=1,2 ..., K, is known parameters for concrete power system；

Step 2: all kinds of parameters required in model in obtaining step 1, including: the bearing power P that in electrical network, system is total_D, computing network loss power parameter B_ij、B_0iAnd B₀₀, the minimum and maximum active power of output of each electromotor in thermal power generation unitWithGenerating set cost parameters a_k、b_kAnd c_i, generating set pollutant discharge amount parameter alpha_k、β_k、γ_k、ξ_kAnd λ_k, i=1,2 ..., K, j=1,2 ..., K, k=1,2 ..., K；

Step 3: adopt multiple target differential evolution algorithm that the mathematical model of step 1 is optimized and solve, it is thus achieved that Pareto disaggregationN is the number solving and concentrating scheduling scheme, solves and concentrates P^g,iRepresent i-th kind of scheduling scheme, i=1,2 ..., N；

Step 4: on the Pareto disaggregation basis that step 3 obtains, utilizes Fuzzy Set Theory method to calculate the satisfaction of each scheduling scheme, and its calculation expression is:

${\mathrm{\&mu;}}_n=\frac{{\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E}{\underset{n=1}{\overset{N}{\&Sigma;}}({\mathrm{\&mu;}}_n^F+{\mathrm{\&mu;}}_n^E)},$

Wherein, μ_nRepresent Pareto solution and concentrate the total satisfaction of the n-th scheduling scheme,Represent the satisfaction of the goal of cost in the n-th scheduling scheme,Represent the satisfaction of pollutant discharge amount in the n-th scheduling scheme, n=1,2 ..., N；WithCan be calculated by following formula respectively:

${\mathrm{\&mu;}}_n^F=\left\{\begin{array}{c}1,F_n\&le;F_{min}\\ \frac{F_{max}-F_n}{F_{max}-F_{\min }},F_{min}<F_n<F_{max}\\ 0,F_n\&GreaterEqual;F_{\max }\end{array}\right.,$

${\mathrm{\&mu;}}_n^E=\left\{\begin{array}{c}1,E_n\&le;E_{min}\\ \frac{E_{max}-E_n}{E_{max}-E_{min}},E_{min}<E_n<E_{max}\\ 0,E_n\&GreaterEqual;E_{max}\end{array}\right.,$

Wherein, F_nAnd E_nRepresent the n-th scheduling scheme generating expense and pollutant discharge amount, F respectively_minAnd F_maxRepresent minima and maximum, the E of the expense that generates electricity in all N number of scheduling schemes respectively_minAnd E_maxRepresent minima and the maximum of pollutant discharge amount in all N number of scheduling schemes respectively；

Step 5: the scheduling scheme satisfaction that step 4 is obtainedNumerical value compare sequence, therefrom select the maximum scheduling scheme of satisfaction numerical value, as final scheduling scheme, and selected scheduling scheme numerical value corresponded in the Pareto forward position scattergram drawn according to this method.

2. a kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm according to claim 1, it is characterized in that, the acquired all kinds of model parameters of described step 2 depend on concrete thermal power plant's generating set, for some thermal power generation unit, if being left out exponential term ξ therein when calculating regulation goal functional value_kexp(λ_kP_k), then can make ξ_kAnd λ_kIt is 0, j=1,2 ..., K, k=1,2 ..., K, this step utilizes the parameter obtained that concrete thermal power plant is carried out Model instantiation.

3. a kind of thermal power plant's environmental economy dispatching method based on multiple target differential evolution algorithm according to claim 1, it is characterised in that described step 3 comprises the following steps:

Step 3.1, initiation parameter NP, W, C, G, α, β；Iteration count initial value g=1 and q is set_l=0, n_l=0, l=1,2,3；Initialize the disaggregation being made up of NP initial solution, also referred to as candidate solution, be designated as P^g={ P^g,1,P^g,2,···,P^g,NP, each initial solutionRepresent a kind of scheduling scheme,Represent the active power of output of kth platform electromotor in i-th scheduling scheme, time initial?In randomly choose, i=1,2 ..., NP, k=1,2 ..., K；

Step 3.2, candidate solution are likely to be unsatisfactory for power-balance constraint, and the solution for not satisfying the constraint condition adopts heuristic to be modified, and revise step and are: for candidate solutionI=1,2 ..., NP, first with B Y-factor method Y computing network lossAnd calculate this candidate solution constraint amount of running counter to according to following formula:

$P_v^{g,i}=P_D+P_L^{g,i}-\underset{k=1}{\overset{K}{\&Sigma;}}{P}_k^{g,i},$

?When being not zero, randomly choose from K electromotor an electromotor k ∈ 1,2 ..., K}, output it powerCurrency increasesThen right according to generator capacity constraintCarry out bound constrained process, be namely set to spanIn boundary value or random value；Recalculate via net loss on this basis

If revised candidate solution still runs counter to power-balance constraint, then repeat above-mentioned makeover process, until it reaches certain reparation number of times or P_cvTerminate when value is less than a certain less threshold value；

Step 3.3, first calculating candidate scheme P^g,iIn the expense of each electromotorAnd pollutant discharge amountK=1,2 ..., K；Then time scheme P is calculated^g,iTotal costAnd gross contamination emissionI=1,2 ..., NP；

Step 3.4, for disaggregation P^gIn i-th candidate solution P^g,i, from the 3 of differential evolution algorithm kinds of mutation operators, { 1,2,3} to P for selection opertor r ∈^g,iCarry out mutation operation, produce variable V^g,i, update the access times n of r operator in nearest W mutation operation simultaneously_r, 3 kinds of differential evolution operators are respectively depicted as:

V^g,i=P^g,r1+α·(P^g,r1-P^g,r1),

V^g,i=P^g,r1+α·(P^g,r2-P^g,r3)+α·(P^g,r4-P^g,r5),

V^g,i=P^g,i+rand(0,1)·(P^g,r1-P^g,i)+α·(P^g,r2-P^g,r3),

In formula, α is the parameter performing corresponding computing, P^g,r1、P^g,r2、P^g,r3、P^g,r4And P^g,r5For participating in the candidate solution of mutation operator in differential evolution operator, they are from disaggregation P^gIn randomly choose, but different, rand (0,1) is the random number between 0 and 1；Wherein, the system of selection of operator r is: if existed in 3 operators from the operator being not used by, then randomly choosing one from the operator being not used by, if 3 operators all used, then selecting by following formula:

$r=\underset{l=1,2,3}{\mathrm{argmax}}(q_l+C\sqrt{\frac{2log\left({\mathrm{\&Sigma;}}_{l=1}^3{n}_l\right)}{n_l}}),$

Wherein, n_lFor the access times of l operator, q in nearest W mutation operation_lFor the accumulation performance of operator l in nearest W mutation operation, C is constant；If P^g,iAdopt the 1st kind or the 2nd kind of mutation operator carries out computing, then proceed as follows further: first produce random integers k_rand∈ 1,2 ..., K}, then for P^g,iIn each variablePerform the intersection operation described by following formula, produce new candidate solution U^g,i:

Wherein,WithRepresent variable V respectively^g,iAnd U^g,iIn kth variable, β for intersect operation desired parameters；If P^g,iThe 3rd kind of operator is adopted to carry out computing, then by new candidate solution U^g,iValue is V^g,i；

If step 3.5 candidate solution U^g,iIt is unsatisfactory for generator capacity constraint and power-balance constraint, then adopts the method described by step 3.2 to be modified, calculate revised candidate solution U^g,iCorresponding cost value and pollutant discharge amount；Then by U^g,iJoin disaggregation P^gIn, now P^gInterior candidate solution number is NP+1；

Step 3.6, according to new disaggregation P^gIn the performance of each candidate solution, delete 1 worst solution, make population number maintain NP；Delete step is: carry out non-dominated ranking according to the cost value of all candidate solutions and pollutant discharge amount, by P^gIt is divided into F₁,F₂,...,F_LIn forward position；If forward position number of plies L is more than 1 and F_LIn only have 1 solution, then directly delete this solution；If forward position number of plies F_LMore than 1 and F_LIn have multiple solution, then delete F_LThe middle solution maximum by other candidate solution domination number of times；If forward position number of plies L is equal to 1, then delete each candidate solution and there is the solution that hypervolume contribution value is minimum, candidate solution hypervolume contribution amount is defined as the area in all candidate solutions and the enclosed region of reference point and deducts and reject the area in residue candidate solution and the enclosed region of reference point after this candidate solution, and wherein reference point each desired value value is the maximum of this desired value in all candidate solutions；

Step 3.7, calculation procedure 3.4 are used the operator effect γ after the r mutation operator_r, and update the accumulation performance q of each operator in nearest W mutation operation_l, l=1,2,3, γ_rEmploying following formula calculates:

γ_r=1/m₁·1/(m₂+1)+m₃,

Wherein, m₁For the candidate solution U using the r mutation operator to produce in step 3.4^g,iForward position sequence number residing during non-bad sequence in step 3.6, m₂For the candidate solution U using the r mutation operator to produce in step 3.4^g,iThe number of times arranged by other candidate solution；m₃For candidate solution U^g,iHypervolume contribution amount；Accumulation performance q_lComputational methods be: for nearest all W mutation operations, W effect value is ranked up from small to large, remembers that the w effect is ordered as R_w, the delay ranking value DR of w effect is then calculated according to following formula_w, it may be assumed that

DR_w=(W-R_w+1)·D^W-w,

The DR calculated after using according to W operator_wValue calculates the AUC area under curve of each operator further, calculates each operator accumulation performance q according to following formula on this basis_l, l=1,2,3:

$q_l=\frac{{\mathrm{AUC}}_l}{{\mathrm{\&Sigma;}}_{l=1}^3{\mathrm{AUC}}_l},$

Wherein, AUC_lIt it is the area in the enclosed region of AUC curve of l operator；

Step 3.8, repetition step 3.4～step 3.7, until P^gIn each candidate solution carried out mutation operation；

Step 3.9, iterations enumerator g value add 1, repeat step 3.4～step 3.8, until iterations enumerator g value reaches maximum iteration time G；Then to disaggregation P^gCarry out non-bad sequence, P^gIn all non-domination solution constitute Pareto disaggregationN is disaggregationThe number of middle solution.