CN104217255A - Electrical power system multi-target overhaul optimization method under market environment - Google Patents

Abstract

The invention discloses an electrical power system multi-target overhaul optimization method under market environment. The electrical power system multi-target overhaul optimization method comprises the following steps: obtaining the data of each electricity generator; establishing a power grid multi-target overhaul optimization model under the market environment; carrying out real number encoding to a unit output variable and a unit overhaul variable of the model, carrying out 0-1 binary encoding to a unit on-line state variable and a unit starting state variable in the model, and converting the unit on-line state variable and the unit starting state variable into dependent variables represented by the unit output variable and the unit overhaul variable from independent variables; initializing the unit output variable and the unit overhaul variable; taking obtained variable initialization values as the population initialization input of a quick non-dominated sorting method for solving to obtain an optimal solution set; and determining a final unit overhaul and output scheme from the obtained optimal solution set by adopting a multi-target decision method. The electrical power system multi-target overhaul optimization method is simple to execute, is high in expandability and can be used for solving multi-target overhaul optimizations of different target functions and constraint conditions.

Description

Electric system multiple goal maintenance optimization method under a kind of market environment

Technical field

The invention belongs to Optimal Technology of Power Systems field, be specifically related to a kind of electric system multiple goal maintenance optimization method.

Background technology

Under Power Market, the arrangement of GENERATOR MAINTENANCE SCHEDULING IN and traditional generator maintenance plan have very large different.Under Power Market, each Power Generation is pursued the maximization of own interests, and meanwhile, system travelling mechanism, as power-management centre, need to ensure the safe and reliable of system.Power Generation wishes that the unit of oneself is arranged in to the low electricity price period to be overhauled, system travelling mechanism wishes to arrange unit maintenance in the underload period, due to low electricity price period and underload period not quite identical, so Power Generation and system travelling mechanism are arranging to exist conflict relationship aspect the unit maintenance period; On the other hand, maximum maintenance unit number and the maintenance capacity that can arrange due to the underload period are limited, so the interests between each Power Generation also exist conflict relationship.Existing turnaround plan adjustment mechanism is generally: each Power Generation is submitted turnaround plan scheme to system travelling mechanism; system travelling mechanism is by certain mechanism; as excitation/penalty mechanism; wish payments mechanisms etc., revise the turnaround plan of Power Generation to reach security of system trading off reliably and between Power Generation interests.This mechanism can be reached one allows each side, containing Power Generation and system travelling mechanism, all satisfied maintenance schemes, but has ignored the research to the conflict relationship between each side, could not understand all sidedly the relation between each side.

Summary of the invention

The object of the invention is for the deficiencies in the prior art, electric system multiple goal maintenance optimization method under a kind of market environment is provided.

The technical solution adopted for the present invention to solve the technical problems is as follows:

Step (1) is obtained each Power Generation generating cost coefficient data C _0ij, C _1ij, C _2ij, unit is $/MW; Recondition expense coefficient data unit is $/MW; Unit starting cost data unit is $; Unit capacity data unit is MW; Load data P _d(t, s), unit is MW, and market electricity price data λ (t, s), unit is $/MWh;

Step (2) is set up electrical network multiple goal maintenance Optimized model under market environment;

Step (3) is carried out real coding to unit output variable and the unit maintenance variable of model described in step (2), machine set on line state variable in described model and unit starting state variable are carried out to 0-1 binary coding, then machine set on line state variable and unit starting state variable are transformed into by unit output variable and the represented dependent variable of unit maintenance variable from independent variable;

Step (4) is carried out initialization to unit output variable and unit maintenance variable in multiple goal maintenance Optimized model;

Step (5) adopts the initialization of variable value obtaining in step (4) to input as the initialization of population of quick non-dominated Sorting method, and adopts NSGA-II to solve above-mentioned multiple goal maintenance Optimized model, obtains optimal solution set;

Step (6) adopts Multiobjective Decision Making Method, determines final unit maintenance and the scheme of exerting oneself from the optimal solution set obtaining.

The objective function of the described Optimized model of step (2) comprises following 3 classes: the maximize revenue function of each Power Generation, system reliability maximize function, the system cost minimization function that always generates electricity;

The constraint condition of described Optimized model comprises following 5 classes: system reserve higher than the required minimum backed-up value of system, genset gross capability and system loading balance, overhaul unit number simultaneously and be less than that higher limit, genset are exerted oneself within the scope of its nominal output, unit can not be simultaneously in maintenance and online two states;

The earnings target function representation of i Power Generation is pf (i), and its expression formula is suc as formula shown in (1):

In formula (1), G _irepresent the unit set of i Power Generation; represent that the j platform unit of i Power Generation is in meritorious the exerting oneself of period t sub-period s; T (t, s) represents that the time of period t sub-period s is long, and unit is hour; y _ij(t, s) represents the starting state of unit, if y _ij(t, s)=1 represents that the j platform unit of i Power Generation started in the zero hour of period t sub-period s, if y _ij(t, s)=0 nothing starts; x _ijthe j platform unit maintenance that represents i Power Generation starts week; D _ijthe j platform unit that represents i Power Generation overhauls duration continuously, and unit is week; represent the max cap. of the j platform unit of i Power Generation; ∨ presentation logic or; Wherein i and j are natural number; M is integer; T represents total time hop count; N represents total period of the day from 11 p.m. to 1 a.m hop count;

Reliability index I (the t of period t sub-period s, s) be expressed as clean for subsequent use for subsequent use divided by hair, the hair capacity by all units for subsequent use and deduct system loading and obtain, clean for subsequent usely deduct the capacity of unit in maintenance and obtain by hair is for subsequent use, shown in (2), system reliability goal function is averaged and is obtained by the reliability index I (t, s) of all sub-periods, shown in (3);

$\max \mathrm{imize}:\frac{1}{T\×N}\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\underset{s=1}{\overset{N}{\mathrm{\Σ}}}I(t,s)---\left(3\right)$

In formula (2), P _d(t, s) represents the system loading of period t sub-period s;

The system the goal of cost function representation that always generates electricity is tc, and its expression formula is suc as formula shown in (4):

System reserve constraint condition is suc as formula shown in (5):

In formula (5), R ^min(t, s) represents that the required minimum of period t sub-period s system is for subsequent use;

Maximum is overhauled unit simultaneously and is counted constraint condition suc as formula shown in (6):

In formula (6), N _i(t) maximum that i Power Generation of expression allows at period t is overhauled unit number simultaneously;

Unit output constraint condition is suc as formula shown in (7):

$v_{\mathrm{ij}}(t,s)P_{G_{\mathrm{ij}}}^{\min }\≤P_{G_{\mathrm{ij}}}(t,s)\≤v_{\mathrm{ij}}(t,s)P_{G_{\mathrm{ij}}}^{\max },\∀t,\∀s---\left(7\right)$

In formula (7), represent the lower limit of exerting oneself when unit is online, v _ij(t, s) is illustrated in line state variable, is 1 online, is not 0 online;

Can not on-line constraints when unit maintenance suc as formula shown in (8):

System power equilibrium constraint is suc as formula shown in (9):

$\underset{i=1}{\overset{I}{\mathrm{\Σ}}}\underset{j\∈G_i}{\mathrm{\Σ}}{P}_{G_{\mathrm{ij}}}(t,s)v_{\mathrm{ij}}(t,s)=P_D(t,s),\∀t,\∀s---\left(9\right).$

Step (3) specifically comprises the steps:

3-1. unit output variable adopt real coding;

3-2. unit maintenance variable x _ijadopt real coding, then round x _ijthe j platform unit maintenance that represents i Power Generation starts week, and maintenance continues duration by D _ijrepresent;

3-3. machine set on line state variable v _ij(t, s) represented by unit output variable, transforms into dependent variable by independent variable, shown in (10):

3-4. unit starting state variable y _ij(t, s) can be by presence variable v _ij(t, s) represents, transforms into dependent variable by independent variable, shown in (11):

y _ij(t+1,1)＝v _ij(t+1,1)-v _ij(t,N)

(11)。

y _ij(t,s+1)＝v _ij(t,s+1)-v _ij(t,s)

Initialization scheme described in step (4) is as follows:

4-1. unit maintenance initialization subroutine is made up of following 6 steps:

4-1-1. input data np and gidx, make t=0, j=0; The unit that wherein np is all Power Generations is counted sum, and array gidx is unit according to the maximum capacity unit number order obtaining that sorts from big to small;

4-1-2. hypothesis unit gidx[j] in maintenance, judge that the maximum shown in the Reserve Constraint shown in formula (5) and formula (6) overhauls the constraint of unit number simultaneously and whether meet simultaneously; If formula (5) and formula (6) are all set up simultaneously, make unit gidx[j] overhaul; If formula (5) and formula (6) have at least one to be false, make unit gidx[j] do not overhaul;

4-1-3. makes j=j+1;

4-1-4. judges whether j<np sets up, if be false redirect execution step 4-1-6; Make t=t+1 if set up;

4-1-5. judges whether t<T sets up, and performs step 4-1-2 if set up; Perform step 4-1-6 if be false; Wherein T is total all numbers that turnaround plan is considered;

4-1-6. returns to sub-maintenance scheme, finishes the initial subroutine of maintenance;

4-2. unit output initialization subroutine is made up of following 6 steps:

It is 0 that 4-2-1. makes all unit outputs, makes j=0;

4-2-2. makes k=0;

4-2-3. makes a decision, if unit pinc[k] in maintenance, perform step 4-2-4; If unit pinc[k] not in maintenance, perform step 4-2-5;

4-2-4. makes k=k+1; Make a decision, if k<ng sets up redirect execution step 4-2-3, perform step 4-2-6 if k<ng is false;

4-2-5. hypothesis unit pinc[k] exert oneself in its upper limit of exerting oneself, make a decision, if all unit outputs and be greater than load reduce unit pinc[k] the power-balance constraint shown in formula (9) is met, execution step 4-2-6; If all unit outputs and be not more than load, make unit pinc[k] in its upper limit of exerting oneself, execution step 4-2-4;

4-2-6.j=j+1; Make a decision, if j<nday sets up, redirect execution step 4-2-2; If j<nday is false, return to unit output scheme, finish unit output initialization subroutine;

Wherein, nday is total number of days that turnaround plan is considered, its value is multiplied by 7 for T; Array pinc is unit according to the average power consumption values of the unit unit number order obtaining that sorts from small to large;

4-3. obtains unit maintenance initialization scheme and unit output initialization scheme is made up of following 7 steps:

4-3-1. carries out permutation and combination to 1 to I, and permutation and combination scheme number is total kind, each permutation and combination scheme execution step 4-3-2 is arrived to step 4-3-6;

Permutation and combination scheme is designated as (i by 4-3-2. ₁, i ₂..., i _k..., i _i), i _krepresent i _kindividual Power Generation;

4-3-3. makes k=1;

4-3-4. note i _kthe unit number that individual Power Generation has is np, and these units are sorted from big to small according to unit maximum capacity, and the unit number order obtaining is placed in array gidx; Call unit maintenance initialization subroutine, the unit maintenance initialization scheme that record obtains;

4-3-5. makes k=k+1; Make a decision, if k≤I, redirect execution step 4-3-4; If k > is I, perform step 4-3-6;

The I that 4-3-6. obtains step 4-3-4 unit maintenance initialization scheme merges the maintenance initialization scheme that obtains all units;

4-3-7. calls unit output initialization subroutine, obtains unit output initialization scheme.

The quick non-dominated Sorting method that step (5) is described, comprises following 7 steps:

5-1. generate initialization population, in population, number of individuals is N _pop, this initialization population is made up of two parts, and Part I is described in step (4) plant unit maintenance and the initialization scheme of exerting oneself; Part II generates at random: unit maintenance variable is made as to an integer of the random generation between 1 to T-1, unit output variable is made as to the random real number generating between unit output lower limit and the unit output upper limit;

Current population P for 5-2. _ggenerate progeny population Q by selection, these 3 genetic operators of crossover and mutation _g, merge current population P _gwith progeny population Q _gobtain mixed population R _g=P _g∪ Q _g;

5-3. is to R _gquick non-dominated Sorting method shown in employing table 1 obtains a series of Paretos forward position, is designated as F={F ₁, F ₂, F ₃..., F _n;

5-4. is from R _gin select in the following way N _popindividuality is as population P of future generation _g+1if: Pareto forward position F ₁middle element number is less than N _pop, F ₁in all elements all put into P _g+1in; Then more next Pareto forward position F ₂if, set P _g+1∪ F ₂in element number be less than N _pop, F ₂in all elements also all put into P _g+1; By that analogy, until there is a Pareto forward position F _k, wherein k ∈ 1,2,3 ..., n}, makes to gather P _g+1∪ F _kin element number be greater than N _pop, calculating F as shown in table 2 _kin the crowding I of each element, so to crowding I arrange from big to small, rank results is placed in F ' _kin, get F ' _kin front N _pop-| P _g+1| individual element is put into P _g+1in, i.e. P _g+1=P _g+1∪ F ' _k[1:(N _pop-| P _g+1|)];

Population P of future generation for 5-5. _g+1generate progeny population Q of future generation by selection, these 3 genetic operators of crossover and mutation _g+1, merge population P of future generation _g+1with progeny population Q of future generation _g+1, obtain mixed population R of future generation _g+1=P _g+1∪ Q _g+1;

5-6. 5-3 is to 5-5, until reach maximum cycle for circulation execution step;

Population P when 5-7. output maximum cycle _g+1, be optimal solution set;

Wherein, | P _g| represent set P _gthe number of middle element, P _g+1∪ F _irepresent P _g+1with F _iunion;

The pseudo-code table of the quick non-dominated Sorting method program of table 1

In table 1, the explanation of symbol <, is minimised as example with multiple goal, and objective function is the smaller the better: q < p set up condition be, and if only if to any i ∈ 1,2 ..., N _objhave and at least exist a j ∈ 1,2 ..., N _objmake

Table 2 calculates the pseudo-code table of crowding program

The described Multiobjective Decision Making Method of step (6) adopts the sort method that approaches ideal value, is specifically divided into 5 steps:

First 6-1., calculates the weighting decision matrix v of standardization _ij:

v _ij＝ω _i(f _i ⁺-f _ij)/(f _i ⁺-f _i ^-),i＝1,2,3,…Nobj,j＝1,2,3,…J (12)

Wherein, f _ijfor j in optimal solution set i the target function value of separating, n _objfor the number of target in multiple-objection optimization, J is the number of separating in optimal solution set, ${\mathrm{\&omega;}}_i=\frac{1}{N_{\mathrm{obj}}}\underset{j=i}{\overset{N_{\mathrm{obj}}}{\mathrm{\&Sigma;}}}\frac{1}{j},i=\mathrm{1,2,3},...,N_{\mathrm{obj}};$

6-2. calculates respectively ideal point least ideal point $A^{-}=\{v_1^{-},v_2^{-},...,v_N^{-}\},$ Wherein, $v_i^{+}=\underset{j}{\max }{v}_{\mathrm{ij}},v_i^{-}=\underset{j}{\min }{v}_{\mathrm{ij}};$

6-3. calculate respectively the distance B of each optimum solution to ideal point ⁺with to the distance B of ideal point least ^-: $D_j^{+}=\sqrt{{\mathrm{\&Sigma;}}_{i=1}^N{(v_{\mathrm{ij}}-v_i^{+})}^2},D_j^{-}=\sqrt{{\mathrm{\&Sigma;}}_{i=1}^N{(v_{\mathrm{ij}}-v_i^{-})}^2},j=\mathrm{1,2,3},...J;$

6-4. calculates the distance ratio of each optimum solution $R_j=D_j^{-}/(D_j^{-}+D_j^{+}),j=\mathrm{1,2,3},...J;$

6-5. is by R _jmaximum optimum solution is chosen as final unit maintenance and the scheme of exerting oneself.

The invention has the beneficial effects as follows:

The present invention proposes electric system multiple goal maintenance optimization method under a kind of market environment.Maintenance under electricity market is a multi-objective optimization question, if adopt single goal optimized algorithm, need to rule of thumb provide the weight of each target, then try to achieve an optimum solution, by changing the weight of each target, then obtain another optimum solution, so repeatedly calculate and can obtain a series of Paretos (Pareto) optimum solution; And Multipurpose Optimal Method, as NSGA-II, in the time solving multi-objective optimization question without the weighted value of given each target, just can obtain all Pareto optimal solutions by once calculating, be conducive to study the relation between interests and the system reliability of each Power Generation under electricity market, can obtain by last decision-making maintenance and the scheme of exerting oneself that a Ge Ling each side is all satisfied with.The Multipurpose Optimal Method that the present invention proposes can solve Electricity Market group of motors turnaround plan problem well, have the following advantages: (1) unit output variable and unit maintenance variable to described model carries out real coding, and machine set on line state variable and unit starting state variable are transformed into by unit output variable and the represented dependent variable of unit maintenance variable from independent variable, the coding of described variable and conversion processing can meet the constraint condition in model better; (2) initialization of population of NSGA-II being carried out, can more effectively solve described multiple goal maintenance Optimized model, obtains more excellent feasible solution; (3) without the weight of given each target, just can obtain all Pareto optimal solutions by once calculating; (4) can obtain by last decision-making technique maintenance and the scheme of exerting oneself that a Ge Ling each side is all satisfied with; (5) described method execution is simple, extensibility is strong, can be used for solving the multiple goal maintenance Optimized model of different target function and constraint condition.

Brief description of the drawings

Fig. 1 is unit maintenance and the scheme initialize routine process flow diagram of exerting oneself that the present invention uses.

Fig. 2 is the unit maintenance initialization subroutine process flow diagram that the present invention uses.

Fig. 3 is the unit output initialization subroutine process flow diagram that the present invention uses.

Fig. 4 is the NSGA-II major cycle process flow diagram that the present invention uses.

Embodiment

Below in conjunction with drawings and Examples, the invention will be further described.

Under a kind of market environment, electric system multiple goal maintenance optimization method, specifically comprises the steps:

Step (1) is obtained each Power Generation generating cost coefficient data C _0ij, C _1ij, C _2ij, unit is $/MW; Recondition expense coefficient data unit is $/MW; Unit starting cost data unit is $; Unit capacity data unit is MW; Load data P _d(t, s), unit is MW, and market electricity price data λ (t, s), unit is $/MWh.

Step (2) is set up electrical network multiple goal maintenance Optimized model under market environment;

The objective function of described Optimized model comprises following 3 classes: the maximize revenue function of each Power Generation, system reliability maximize function, the system cost minimization function that always generates electricity;

The constraint condition of described Optimized model comprises following 5 classes: system reserve higher than the required minimum backed-up value of system, genset gross capability and system loading balance, overhaul unit number simultaneously and be less than that higher limit, genset are exerted oneself within the scope of its nominal output, unit can not be simultaneously in maintenance and online two states;

The earnings target function representation of i Power Generation is pf (i), and its expression formula is suc as formula shown in (1):

In formula (1), G _irepresent the unit set of i Power Generation; represent that the j platform unit of i Power Generation is in meritorious the exerting oneself of period t sub-period s; T (t, s) represents that the time of period t sub-period s is long, and unit is hour; y _ij(t, s) represents the starting state of unit, if y _ij(t, s)=1 represents that the j platform unit of i Power Generation started in the zero hour of period t sub-period s, if y _ij(t, s)=0 nothing starts; x _ijthe j platform unit maintenance that represents i Power Generation starts week; D _ijthe j platform unit that represents i Power Generation overhauls duration continuously, and unit is week; represent the max cap. of the j platform unit of i Power Generation; ∨ presentation logic or; Wherein i and j are natural number; M is integer; T represents total time hop count; N represents total period of the day from 11 p.m. to 1 a.m hop count.

Reliability index I (the t of period t sub-period s, s) be expressed as clean for subsequent use for subsequent use divided by hair, the hair capacity by all units for subsequent use and deduct system loading and obtain, clean for subsequent usely deduct the capacity of unit in maintenance and obtain by hair is for subsequent use, shown in (2), system reliability goal function is averaged and is obtained by the reliability index I (t, s) of all sub-periods, shown in (3).

$\max \mathrm{imize}:\frac{1}{T\&times;N}\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}\underset{s=1}{\overset{N}{\mathrm{\&Sigma;}}}I(t,s)---\left(3\right)$

In formula (2), P _d(t, s) represents the system loading of period t sub-period s.

The system the goal of cost function representation that always generates electricity is tc, and its expression formula is suc as formula shown in (4):

System reserve constraint condition is suc as formula shown in (5):

In formula (5), R ^min(t, s) represents that the required minimum of period t sub-period s system is for subsequent use.

Maximum is overhauled unit simultaneously and is counted constraint condition suc as formula shown in (6):

In formula (6), N _i(t) maximum that i Power Generation of expression allows at period t is overhauled unit number simultaneously.

Unit output constraint condition is suc as formula shown in (7):

$v_{\mathrm{ij}}(t,s)P_{G_{\mathrm{ij}}}^{\min }\&le;P_{G_{\mathrm{ij}}}(t,s)\&le;v_{\mathrm{ij}}(t,s)P_{G_{\mathrm{ij}}}^{\max },\&ForAll;t,\&ForAll;s---\left(7\right)$

In formula (7), represent the lower limit of exerting oneself when unit is online, v _ij(t, s) is illustrated in line state variable, is 1 online, is not 0 online;

Can not on-line constraints when unit maintenance suc as formula shown in (8):

System power equilibrium constraint is suc as formula shown in (9):

$\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}\underset{j\&Element;G_i}{\mathrm{\&Sigma;}}{P}_{G_{\mathrm{ij}}}(t,s)v_{\mathrm{ij}}(t,s)=P_D(t,s),\&ForAll;t,\&ForAll;s---\left(9\right)$

Step (3) is carried out real coding to unit output variable and the unit maintenance variable of model described in step (2), machine set on line state variable in described model and unit starting state variable are carried out to 0-1 binary coding, then machine set on line state variable and unit starting state variable are transformed into by unit output variable and the represented dependent variable of unit maintenance variable from independent variable; Concrete steps are as follows:

3-1. unit output variable adopt real coding;

3-2. unit maintenance variable x _ijadopt real coding, then round x _ijthe j platform unit maintenance that represents i Power Generation starts week, and maintenance continues duration by D _ijrepresent;

3-3. machine set on line state variable v _ij(t, s) represented by unit output variable, transforms into dependent variable by independent variable, shown in (10):

3-4. unit starting state variable y _ij(t, s) can be by presence variable v _ij(t, s) represents, transforms into dependent variable by independent variable, shown in (11):

y _ij(t+1,1)＝v _ij(t+1,1)-v _ij(t,N)

(11)

y _ij(t,s+1)＝v _ij(t,s+1)-v _ij(t,s)

Step (4) is carried out initialization to unit output variable and unit maintenance variable in multiple goal maintenance Optimized model.Initialization scheme is as follows:

Here first describe unit maintenance initialization subroutine, then describe unit output initialization subroutine, finally describe and utilize the first two subroutine call to unit maintenance initialization scheme and unit output initialization scheme.

As shown in Figure 1, unit maintenance initialization subroutine is made up of following 6 steps:

4-1-1. input data np and gidx, make t=0, j=0; The unit that wherein np is all Power Generations is counted sum, and array gidx is unit according to the maximum capacity unit number order obtaining that sorts from big to small;

4-1-2. hypothesis unit gidx[j] in maintenance, judge that the maximum shown in the Reserve Constraint shown in formula (5) and formula (6) overhauls the constraint of unit number simultaneously and whether meet simultaneously; If formula (5) and formula (6) are all set up simultaneously, make unit gidx[j] overhaul; If formula (5) and formula (6) have at least one to be false, make unit gidx[j] do not overhaul;

4-1-3. makes j=j+1;

4-1-4. judges whether j<np sets up, if be false redirect execution step 4-1-6; Make t=t+1 if set up;

4-1-5. judges whether t<T sets up, and performs step 4-1-2 if set up; Perform step 4-1-6 if be false; Wherein T is total all numbers that turnaround plan is considered;

4-1-6. returns to sub-maintenance scheme, finishes the initial subroutine of maintenance;

As shown in Figure 2, unit output initialization subroutine is made up of following 6 steps:

It is 0 that 4-2-1. makes all unit outputs, makes j=0;

4-2-2. makes k=0;

4-2-3. makes a decision, if unit pinc[k] in maintenance, perform step 4-2-4; If unit pinc[k] not in maintenance, perform step 4-2-5;

4-2-4. makes k=k+1; Make a decision, if k<ng sets up redirect execution step 4-2-3, perform step 4-2-6 if k<ng is false;

4-2-5. hypothesis unit pinc[k] exert oneself in its upper limit of exerting oneself, make a decision, if all unit outputs and be greater than load reduce unit pinc[k] the power-balance constraint shown in formula (9) is met, execution step 4-2-6; If all unit outputs and be not more than load, make unit pinc[k] in its upper limit of exerting oneself, execution step 4-2-4;

4-2-6.j=j+1; Make a decision, if j<nday sets up, redirect execution step 4-2-2; If j<nday is false, return to unit output scheme, finish unit output initialization subroutine;

Wherein, nday is total number of days that turnaround plan is considered, its value is multiplied by 7 for T; Array pinc is unit according to the average power consumption values of the unit unit number order obtaining that sorts from small to large.

As shown in Figure 3, the method that obtains unit maintenance initialization scheme and unit output initialization scheme is made up of following 7 steps:

4-3-1. carries out permutation and combination to 1 to I, and permutation and combination scheme number is total kind, each permutation and combination scheme execution step 4-3-2 is arrived to step 4-3-6;

Permutation and combination scheme is designated as (i by 4-3-2. ₁, i ₂..., i _k..., i _i), i _krepresent i _kindividual Power Generation;

4-3-3. makes k=1;

4-3-4. note i _kthe unit number that individual Power Generation has is np, and these units are sorted from big to small according to unit maximum capacity, and the unit number order obtaining is placed in array gidx; Call unit maintenance initialization subroutine, the unit maintenance initialization scheme that record obtains;

4-3-5. makes k=k+1; Make a decision, if k≤I, redirect execution step 4-3-4; If k > is I, perform step 4-3-6;

The I that 4-3-6. obtains step 4-3-4 unit maintenance initialization scheme merges the maintenance initialization scheme that obtains all units;

4-3-7. calls unit output initialization subroutine, obtains unit output initialization scheme.

Unit maintenance shown in Fig. 3 and the scheme initialize routine of exerting oneself have successively been called the unit output initialization subroutine shown in the unit maintenance initialization subroutine shown in Fig. 1 and Fig. 2.

Step (5) adopts the initialization of variable value obtaining in step (4) to input as the initialization of population of quick non-dominated Sorting method (NSGA-II), and adopt NSGA-II to solve above-mentioned multiple goal maintenance Optimized model, obtain optimal solution set:

As shown in Figure 4, NSGA-II comprises following 7 steps:

5-1. generate initialization population, in population, number of individuals is N _pop, this initialization population is made up of two parts, and Part I is described in step (4) plant unit maintenance and the initialization scheme of exerting oneself; Part II generates at random: unit maintenance variable is made as to an integer of the random generation between 1 to T-1, unit output variable is made as to the random real number generating between unit output lower limit and the unit output upper limit.

Current population P for 5-2. _ggenerate progeny population Q by selection, these 3 genetic operators of crossover and mutation _g, merge current population P _gwith progeny population Q _gobtain mixed population R _g=P _g∪ Q _g.

5-3. is to R _gquick non-dominated Sorting method shown in employing table 1 obtains a series of Paretos forward position (Pareto front), is designated as F={F ₁, F ₂, F ₃..., F _n.

5-4. is from R _gin select in the following way N _popindividuality is as population P of future generation _g+1if: Pareto forward position F ₁middle element number is less than N _pop, F ₁in all elements all put into P _g+1in; Then more next Pareto forward position F ₂if, set P _g+1∪ F ₂in element number be less than N _pop, F ₂in all elements also all put into P _g+1; By that analogy, until there is a Pareto forward position F _k, wherein k ∈ 1,2,3 ..., n}, makes to gather P _g+1∪ F _kin element number be greater than N _pop, calculating F as shown in table 2 _kin the crowding I of each element, so to crowding I arrange from big to small, rank results is placed in F ' _kin, get F ' _kin front N _pop-| P _g+1| individual element is put into P _g+1in, i.e. P _g+1=P _g+1∪ F ' _k[1:(N _pop-| P _g+1|)].

Population P of future generation for 5-5. _g+1generate progeny population Q of future generation by selection, these 3 genetic operators of crossover and mutation _g+1, merge population P of future generation _g+1with progeny population Q of future generation _g+1, obtain mixed population R of future generation _g+1=P _g+1∪ Q _g+1.

5-6. 5-3 is to 5-5, until reach maximum cycle for circulation execution step.

Population P when 5-7. output maximum cycle _g+1, be optimal solution set.

Wherein, | P _g| represent set P _gthe number of middle element, P _g+1∪ F _irepresent P _g+1with F _iunion.

The pseudo-code table of the quick non-dominated Sorting method program of table 1

In table 1, the explanation of symbol <, is minimised as example with multiple goal, and objective function is the smaller the better: q < p set up condition be, and if only if to any i ∈ 1,2 ..., N _objhave and at least exist a j ∈ 1,2 ..., N _objmake

Table 2 calculates the pseudo-code table of crowding program

Step (6) adopts Multiobjective Decision Making Method, determines final unit maintenance and the scheme of exerting oneself from the optimal solution set obtaining.Described Multiobjective Decision Making Method adopts the sort method (Technique for Order Preference by Similarity to Ideal Solution, TOPSIS) that approaches ideal value.TOPSIS can be divided into 5 steps:

First 6-1., calculates the weighting decision matrix vi of standardization _j:

v _ij＝ω _i(f _i ⁺-f _ij)/(f _i ⁺-f _i ^-),i＝1,2,3,…Nobj,j＝1,2,3,…J (12)

Wherein, f _ijfor j in optimal solution set i the target function value of separating, n _objfor the number of target in multiple-objection optimization, J is the number of separating in optimal solution set, ${\mathrm{\&omega;}}_i=\frac{1}{N_{\mathrm{obj}}}\underset{j=i}{\overset{N_{\mathrm{obj}}}{\mathrm{\&Sigma;}}}\frac{1}{j},i=\mathrm{1,2,3},...,N_{\mathrm{obj}};$

6-2. calculates respectively ideal point least ideal point $A^{-}=\{v_1^{-},v_2^{-},...,v_N^{-}\},$ Wherein, $v_i^{+}=\underset{j}{\max }{v}_{\mathrm{ij}},v_i^{-}=\underset{j}{\min }{v}_{\mathrm{ij}}.$

6-3. calculate respectively the distance B of each optimum solution to ideal point ⁺with to the distance B of ideal point least ^-: $D_j^{+}=\sqrt{{\mathrm{\&Sigma;}}_{i=1}^N{(v_{\mathrm{ij}}-v_i^{+})}^2},D_j^{-}=\sqrt{{\mathrm{\&Sigma;}}_{i=1}^N{(v_{\mathrm{ij}}-v_i^{-})}^2},j=\mathrm{1,2,3},...J.$

6-4. calculates the distance ratio of each optimum solution $R_j=D_j^{-}/(D_j^{-}+D_j^{+}),j=\mathrm{1,2,3},...J.$

6-5. is by R _jmaximum optimum solution is chosen as final unit maintenance and the scheme of exerting oneself.

Step (7) power-management centre is according to unit maintenance final in step (6) and the scheme of exerting oneself, thereby by information processing and communicator, the inspecting state of every unit and the value of exerting oneself assigned to the running status of DCS of Power Plant automatic generator group and regulated it to exert oneself.

First briefly introduce two concepts for following narration is convenient: DCS of Power Plant and Automatic Generation Control.DCS of Power Plant (Distributed Control System, DCS) be taking microcomputer as basis, according to the concept of system control, merged computer technology, control technology, the communication technology and graphical display technics, realize centralized management, decentralised control.DCS system has become the major equipment that power plant controls, monitors.Automatic Generation Control (Automatic Generation Control, AGC), it is the important component part of energy management system.By the control target of grid dispatching center, instruction is sent to concerned power generation factory or unit, by the automatic control and adjustment device of power plant or unit, realize the automatic control to the power of the assembling unit.

In Network of Power supervisory system, often adopt information processing and communicator D200.Up to the present, exceed 200 power plant in the whole nation, comprise that exceeding 65% large power plant adopts D200 to realize scheduling the AGC of power plant is controlled and remote moving function.Comprising the power plant of numerous 600MW units, such as Zhejiang Jiaxing power plant, Ninghai power plant, Sanmenxia Gorge power plant etc.AGC order can directly be sent out to DCS system by information processing and communicator D200 in power-management centre, to realize, unit output is regulated.

Power-management centre is according to unit maintenance final in step (6) and the scheme of exerting oneself, thereby by information processing and communicator D200, the inspecting state of every unit and the value of exerting oneself assigned to the running status of DCS system automatic generator group and regulated it to exert oneself.If unit needs maintenance; power-management centre is assigned computer supervisory control system to generating plant by auto stop signal with AGC order by information processing and communicator D200; power plant's computer supervisory control system receives after scheduling AGC order, sends the instruction of shutting down maintenance to unit.If unit does not need maintenance, power-management centre can directly be assigned DCS system to genset by the value of exerting oneself of every unit with the form of AGC order by information processing and communicator D200, to realize the automatic control that genset is exerted oneself.Specific as follows: power-management centre obtains every genset value of exerting oneself, by information processing and communicator D200 with AGC instruction issuing to each generating plant; The host CPU plate of generating plant D200 first receives the AGC instruction of assigning dispatching center, then through stipulations information processings by the value of AGC instruction number and send to the CPU board of D20C compoboard; The AGC instruction quantizing of receiving is changed into genset 0～100% corresponding code value of exerting oneself by D20C compoboard, and according to the DC analogue quantity of this code value output 4～20mA the DCS system to genset, thereby the exerting oneself of regulator generator group.

Embodiment 1

In this patent hypothesis electricity market, have 3 Power Generations, according to permutation and combination, according to the sequencing difference of determining Power Generation unit maintenance scheme, Ke Yiyou kind, i.e. (1,2,3), (1,3,2), (2,1,3), (2,3,1), (3,1,2) and (3,2,1) these 6 kinds.What (1,2,3) represented is the unit maintenance scheme of first determining Power Generation 1, then determines the unit maintenance scheme of Power Generation 2, finally determines the process flow diagram of the unit maintenance scheme of Power Generation 3.Unit is arranged in to the low electricity price period, and to overhaul its opportunity cost lower, and the unit capacity and the unit sum that overhaul due to each period are limited simultaneously, therefore if first determine the unit maintenance scheme of Power Generation 1, the income of Power Generation 1 will be larger; And if first determine the unit maintenance scheme of Power Generation 2, the income of Power Generation 2 will be larger.In this patent, obtain altogether 6 kinds of unit maintenances and the initialization scheme of exerting oneself.

Embodiment 2

With each Power Generation generating expense of obtaining in electric system under certain Power Market, recondition expense, start and stop expense, unit capacity and following 52 weeks, i.e. T=52, the market weekly data such as average electricity price is data used in the present invention.Other parameter arranges as follows, R ^min(t, s) is taken as 0.05 times of period t sub-period s system loading, N _i(t) be taken as 3, sub-period N=7,3 of Power Generation numbers, 32 of total units, i.e. np=32, objective function number N _objthe maximum iteration time being taken as in 5, NSGA-II is made as 1500, total individual number N in population _popbe made as 1200.

First, in order to verify that the initial method described in the coding transformation processing method described in step in the present invention (3) and step (4) is to using the validity of the Electricity Market group of motors turnaround plan problem described in described NSGA-II Algorithm for Solving, we have been four contrast experiments as shown in table 3, and its solving result is as shown in table 4.

Four contrast experiments' of table 3 setting

Four contrast experiments' of table 4 result

Objective function f1 in table 4, f2 and f3 represent respectively the income of Power Generation 1,2 and 3, and the larger expression scheme of income is better, so objective function f1, f2 and f3 must adopt to maximize and optimize, and minimize optimization, so in NSGA-II solution procedure because NSGA-II can only process, we are by f1, f2 and f3 get negative value, minimize-f1 ,-f2 and-f3.Objective function f5 represents system reliability goal function, this function representation system reserve value, and system reserve is higher shows that system stability is better, optimize therefore objective function f5 also must adopt to maximize, be similar to f1, the processing mode of f2 and f3, f5 is got negative value by we, minimizes-f5.Objective function f4 represents the system expense of always generating electricity, and the expense of always generating electricity is more low better, minimize optimization, without special processing therefore objective function f4 must adopt.

From table 4, we can see, do not adopt the code processing method shown in step (3) in NSGA-II-1 and NSGA-II-2, cannot obtain feasible solution; In NSGA-II-3 and NSGA-II-4, all adopted the code processing method shown in step (3), can successfully obtain optimal solution set, in this optimal solution set, the optimal value of each objective function is as shown in table 4.This shows that the code processing method shown in step (3) plays very important effect for solving smoothly of problem.

NSGA-II-4 is compared with NSGA-II-3, and containing the initialization in step (4), the latter does not contain for the former.Result in table 4 shows the objective function f1 that the former obtains, f2, and the corresponding maximal value that the maximal value of f3 and f5 all obtains than the latter is large; The minimum value that the minimum value of the objective function f4 that the former obtains obtains than the latter is little.This shows that the initialization shown in step (4) is conducive to better be separated.Through checking, solutions all in the optimal solution set that NSGA-II-3 and NSGA-II-4 try to achieve all meet constraint condition, are feasible solution.

For explaining conveniently, final unit maintenance and the scheme of exerting oneself in the optimal solution set that we claim to adopt Multiobjective Decision Making Method TOPSIS to obtain from step (5), determined are separated for final.Table 5 has provided each target function value of final solution, and table 6 has provided the unit maintenance scheme of final solution, and final unit output data volume of separating is larger, has 32 × 364=11648 data, therefore do not provide.In sum, it is effective that a kind of Multipurpose Optimal Method NSAG-II proposing in the present invention is used for solving Electricity Market group of motors turnaround plan.

Final each target function value separated of table 5

	f1/×10 8＄	f2/×10 8＄	f3/×10 8＄	f4/×10 8＄	f5
						NSGA-II-4	1.75	1.35	1.99	5.77	0.8715

The final unit maintenance scheme of separating of table 6

Unit sequence number	The unit output upper limit	Maintenance period/week	Unit sequence number	The unit output upper limit	Maintenance period/week
						1	20	31-32	17	12	41-42
2	20	18-19	18	12	9-10
						3	76	11-12	19	12	42-43
4	76	45-46	20	155	12-13
						5	20	36-37	21	155	46-47
6	20	40-41	22	155	44-45
						7	76	34-35	23	155	35-36
8	76	43-44	24	350	33-34
						9	100	41-42	25	400	51-52
10	100	9-10	26	400	48-49
						11	100	39-40	27	50	51-52
12	197	13-14	28	50	49-50
						13	197	44-45	29	50	33-34
14	197	35-36	30	50	47-48
						15	12	39-40	31	50	13-14
16	12	32-33	32	50	46-47

。

Claims (6)

1. an electric system multiple goal maintenance optimization method under market environment, is characterized in that comprising the steps:

Step (1) is obtained each Power Generation generating cost coefficient data C _0ij, C _1ij, C _2ij, unit is $/MW; Recondition expense coefficient data unit is $/MW; Unit starting cost data unit is $; Unit capacity data unit is MW; Load data P _d(t, s), unit is MW, and market electricity price data λ (t, s), unit is $/MWh;

Step (2) is set up electrical network multiple goal maintenance Optimized model under market environment;

Step (3) is carried out real coding to unit output variable and the unit maintenance variable of model described in step (2), machine set on line state variable in described model and unit starting state variable are carried out to 0-1 binary coding, then machine set on line state variable and unit starting state variable are transformed into by unit output variable and the represented dependent variable of unit maintenance variable from independent variable;

Step (4) is carried out initialization to unit output variable and unit maintenance variable in multiple goal maintenance Optimized model;

Step (5) adopts the initialization of variable value obtaining in step (4) to input as the initialization of population of quick non-dominated Sorting method, and adopts NSGA-II to solve above-mentioned multiple goal maintenance Optimized model, obtains optimal solution set;

Step (6) adopts Multiobjective Decision Making Method, determines final unit maintenance and the scheme of exerting oneself from the optimal solution set obtaining.

2. electric system multiple goal maintenance optimization method under a kind of market environment as claimed in claim 1, is characterized in that the objective function of the described Optimized model of step (2) comprises following 3 classes: the maximize revenue function of each Power Generation, system reliability maximize function, the system cost minimization function that always generates electricity;

The constraint condition of described Optimized model comprises following 5 classes: system reserve higher than the required minimum backed-up value of system, genset gross capability and system loading balance, overhaul unit number simultaneously and be less than that higher limit, genset are exerted oneself within the scope of its nominal output, unit can not be simultaneously in maintenance and online two states;

The earnings target function representation of i Power Generation is pf (i), and its expression formula is suc as formula shown in (1):

In formula (1), G _irepresent the unit set of i Power Generation; represent that the j platform unit of i Power Generation is in meritorious the exerting oneself of period t sub-period s; T (t, s) represents that the time of period t sub-period s is long, and unit is hour; y _ij(t, s) represents the starting state of unit, if y _ij(t, s)=1 represents that the j platform unit of i Power Generation started in the zero hour of period t sub-period s, if y _ij(t, s)=0 nothing starts; x _ijthe j platform unit maintenance that represents i Power Generation starts week; D _ijthe j platform unit that represents i Power Generation overhauls duration continuously, and unit is week; represent the max cap. of the j platform unit of i Power Generation; ∨ presentation logic or; Wherein i and j are natural number; M is integer; T represents total time hop count; N represents total period of the day from 11 p.m. to 1 a.m hop count;

Reliability index I (the t of period t sub-period s, s) be expressed as clean for subsequent use for subsequent use divided by hair, the hair capacity by all units for subsequent use and deduct system loading and obtain, clean for subsequent usely deduct the capacity of unit in maintenance and obtain by hair is for subsequent use, shown in (2), system reliability goal function is averaged and is obtained by the reliability index I (t, s) of all sub-periods, shown in (3);

In formula (2), P _d(t, s) represents the system loading of period t sub-period s;

The system the goal of cost function representation that always generates electricity is tc, and its expression formula is suc as formula shown in (4):

System reserve constraint condition is suc as formula shown in (5):

In formula (5), R ^min(t, s) represents that the required minimum of period t sub-period s system is for subsequent use;

Maximum is overhauled unit simultaneously and is counted constraint condition suc as formula shown in (6):

In formula (6), N _i(t) maximum that i Power Generation of expression allows at period t is overhauled unit number simultaneously;

Unit output constraint condition is suc as formula shown in (7):

In formula (7), represent the lower limit of exerting oneself when unit is online, v _ij(t, s) is illustrated in line state variable, is 1 online, is not 0 online;

Can not on-line constraints when unit maintenance suc as formula shown in (8):

System power equilibrium constraint is suc as formula shown in (9):

3. electric system multiple goal maintenance optimization method under a kind of market environment as claimed in claim 1, is characterized in that step (3) specifically comprises the steps:

3-1. unit output variable adopt real coding;

3-2. unit maintenance variable x _ijadopt real coding, then round x _ijthe j platform unit maintenance that represents i Power Generation starts week, and maintenance continues duration by D _ijrepresent;

3-3. machine set on line state variable v _ij(t, s) represented by unit output variable, transforms into dependent variable by independent variable, shown in (10):

3-4. unit starting state variable y _ij(t, s) can be by presence variable v _ij(t, s) represents, transforms into dependent variable by independent variable, shown in (11):

y _ij(t+1,1)＝v _ij(t+1,1)-v _ij(t,N)

y _ij(t,s+1)＝v _ij(t,s+1)-v _ij(t,s) (11)。

4. electric system multiple goal maintenance optimization method under a kind of market environment as claimed in claim 1, is characterized in that the described initialization scheme of step (4) is as follows:

4-1. unit maintenance initialization subroutine is made up of following 6 steps:

4-1-1. input data np and gidx, make t=0, j=0; The unit that wherein np is all Power Generations is counted sum, and array gidx is unit according to the maximum capacity unit number order obtaining that sorts from big to small;

4-1-2. hypothesis unit gidx[j] in maintenance, judge that the maximum shown in the Reserve Constraint shown in formula (5) and formula (6) overhauls the constraint of unit number simultaneously and whether meet simultaneously; If formula (5) and formula (6) are all set up simultaneously, make unit gidx[j] overhaul; If formula (5) and formula (6) have at least one to be false, make unit gidx[j] do not overhaul;

4-1-3. makes j=j+1;

4-1-4. judges whether j<np sets up, if be false redirect execution step 4-1-6; Make t=t+1 if set up;

4-1-5. judges whether t<T sets up, and performs step 4-1-2 if set up; Perform step 4-1-6 if be false; Wherein T is total all numbers that turnaround plan is considered;

4-1-6. returns to sub-maintenance scheme, finishes the initial subroutine of maintenance;

4-2. unit output initialization subroutine is made up of following 6 steps:

It is 0 that 4-2-1. makes all unit outputs, makes j=0;

4-2-2. makes k=0;

4-2-3. makes a decision, if unit pinc[k] in maintenance, perform step 4-2-4; If unit pinc[k] not in maintenance, perform step 4-2-5;

4-2-4. makes k=k+1; Make a decision, if k<ng sets up redirect execution step 4-2-3, perform step 4-2-6 if k<ng is false;

4-2-5. hypothesis unit pinc[k] exert oneself in its upper limit of exerting oneself, make a decision, if all unit outputs and be greater than load reduce unit pinc[k] the power-balance constraint shown in formula (9) is met, execution step 4-2-6; If all unit outputs and be not more than load, make unit pinc[k] in its upper limit of exerting oneself, execution step 4-2-4;

4-2-6.j=j+1; Make a decision, if j<nday sets up, redirect execution step 4-2-2; If j<nday is false, return to unit output scheme, finish unit output initialization subroutine;

Wherein, nday is total number of days that turnaround plan is considered, its value is multiplied by 7 for T; Array pinc is unit according to the average power consumption values of the unit unit number order obtaining that sorts from small to large;

4-3. obtains unit maintenance initialization scheme and unit output initialization scheme is made up of following 7 steps:

4-3-1. carries out permutation and combination to 1 to I, and permutation and combination scheme number is total kind, each permutation and combination scheme execution step 4-3-2 is arrived to step 4-3-6;

Permutation and combination scheme is designated as (i by 4-3-2. ₁, i ₂..., i _k..., i _i), i _krepresent i _kindividual Power Generation;

4-3-3. makes k=1;

4-3-4. note i _kthe unit number that individual Power Generation has is np, and these units are sorted from big to small according to unit maximum capacity, and the unit number order obtaining is placed in array gidx; Call unit maintenance initialization subroutine, the unit maintenance initialization scheme that record obtains;

4-3-5. makes k=k+1; Make a decision, if k≤I, redirect execution step 4-3-4; If k > is I, perform step 4-3-6;

The I that 4-3-6. obtains step 4-3-4 unit maintenance initialization scheme merges the maintenance initialization scheme that obtains all units;

4-3-7. calls unit output initialization subroutine, obtains unit output initialization scheme.

5. electric system multiple goal maintenance optimization method under a kind of market environment as claimed in claim 1, is characterized in that described quick non-dominated Sorting method, comprises following 7 steps:

5-1. generate initialization population, in population, number of individuals is N _pop, this initialization population is made up of two parts, and Part I is described in step (4) plant unit maintenance and the initialization scheme of exerting oneself; Part II generates at random: unit maintenance variable is made as to an integer of the random generation between 1 to T-1, unit output variable is made as to the random real number generating between unit output lower limit and the unit output upper limit;

Current population P for 5-2. _ggenerate progeny population Q by selection, these 3 genetic operators of crossover and mutation _g, merge current population P _gwith progeny population Q _gobtain mixed population R _g=P _g∪ Q _g;

5-3. is to R _gquick non-dominated Sorting method shown in employing table 1 obtains a series of Paretos forward position, is designated as F={F ₁, F ₂, F ₃..., F _n;

5-4. is from R _gin select in the following way N _popindividuality is as population P of future generation _g+1if: Pareto forward position F ₁middle element number is less than N _pop, F ₁in all elements all put into P _g+1in; Then more next Pareto forward position F ₂if, set P _g+1∪ F ₂in element number be less than N _pop, F ₂in all elements also all put into P _g+1; By that analogy, until there is a Pareto forward position F _k, wherein k ∈ 1,2,3 ..., n}, makes to gather P _g+1∪ F _kin element number be greater than N _pop, calculating F as shown in table 2 _kin the crowding I of each element, so to crowding I arrange from big to small, rank results is placed in F ' _kin, get F ' _kin front N _pop-| P _g+1| individual element is put into P _g+1in, i.e. P _g+1=P _g+1∪ F ' _k[1:(N _pop-| P _g+1|)];

Population P of future generation for 5-5. _g+1generate progeny population Q of future generation by selection, these 3 genetic operators of crossover and mutation _g+1, merge population P of future generation _g+1with progeny population Q of future generation _g+1, obtain mixed population R of future generation _g+1=P _g+1∪ Q _g+1;

5-6. 5-3 is to 5-5, until reach maximum cycle for circulation execution step;

Population P when 5-7. output maximum cycle _g+1, be optimal solution set;

Wherein, | P _g| represent set P _gthe number of middle element, P _g+1∪ F _irepresent P _g+1with F _iunion;

The pseudo-code table of the quick non-dominated Sorting method program of table 1

In table 1, the explanation of symbol <, is minimised as example with multiple goal, and objective function is the smaller the better: q < p set up condition be, and if only if to any i ∈ 1,2 ..., N _objhave and at least exist a j ∈ 1,2 ..., N _objmake

Table 2 calculates the pseudo-code table of crowding program

6. electric system multiple goal maintenance optimization method under a kind of market environment as claimed in claim 1, is characterized in that the described Multiobjective Decision Making Method of step (6) adopts the sort method that approaches ideal value, is specifically divided into 5 steps:

First 6-1., calculates the weighting decision matrix v of standardization _ij:

v _ij＝ω _i(f _i ⁺-f _ij)/(f _i ⁺-f _i ^-),i＝1,2,3,…Nobj,j＝1,2,3,…J (12)

Wherein, f _ijfor j in optimal solution set i the target function value of separating, n _objfor the number of target in multiple-objection optimization, J is the number of separating in optimal solution set,

6-2. calculates respectively ideal point least ideal point wherein,

6-3. calculate respectively the distance B of each optimum solution to ideal point ⁺with to the distance B of ideal point least ^-:

6-4. calculates the distance ratio of each optimum solution

6-5. is by R _jmaximum optimum solution is chosen as final unit maintenance and the scheme of exerting oneself.