CN104600713A - Device and method for generating day-ahead reactive power dispatch of power distribution network containing wind/photovoltaic power generation - Google Patents

Abstract

The invention discloses a device for generating day-ahead reactive power dispatch of a power distribution network containing wind/photovoltaic power generation. The device comprises a scene optimization dispatching base off-line formation module for clustering and combining wind/photovoltaic power generation historical data to form a scene set and calculating the dispatching scheme of each scene to form an optimized scene base, a next-day state mode matching module for performing mode matching according to the predicted 24-hour state of the next day to form a next-day dispatching scheme solution set, a capacitor switching scheme correction module for correcting a next-day 24-hour capacitor switching scheme, and a multi-attribute decision-making module for obtaining the optimal scheme and a second-best scheme for the reference of the dispatching personnel. The invention also provides a method utilizing the device.

Description

Containing the wind power/photovoltaic generation distribution net generating apparatus of Reactive Power Dispatch and method a few days ago

Technical field

The invention belongs to the technical field of Automation of Electric Systems, relate to a kind of containing the wind power/photovoltaic generation distribution net generating apparatus of Reactive Power Dispatch and method a few days ago particularly.

Background technology

Distributed power source has inexhaustible and feature that is cleanliness without any pollution, effectively supplementing as centralized generating, its access power distribution network becomes inexorable trend, and the distributed power source therefore based on regenerative resource is greatly developed in each provinces and cities' grid company, studied and Demonstration Application.Recent research document shows that distributed power source can not only provide active power, also can provide reactive power support for system, participates in power distribution network Reactive Power Dispatch, and it is idle, and size of exerting oneself depends on gain merit force value and inverter capacity.But the prediction of exerting oneself of distributed power source is more difficult, usually cannot predict accurately on a larger time scale, is therefore difficult to perform to its Reactive Power Dispatch; Higher to the precision of prediction of distributed power source in time scale a few days ago, but the time leaving traffic department's Optimized Operation for is shorter, and a large amount of on-line search formula optimized algorithms can not meet temporal requirement usually.Therefore the power distribution network Reactive Power Dispatch device simultaneously meeting distributed power source precision of prediction and rapid Optimum scheduling is not realized so far.Existing research is the idle work optimization in time scale a few days ago mostly, but the used time is longer, and the actual track for complexity is inapplicable.

Therefore, for strengthening the utilization to distributed power source further, the more reasonably all types of reactive power source of coordinated scheduling power distribution network, fast and promote electrical network index and economic index to greatest extent, needs to propose a kind of containing the wind power/photovoltaic generation distribution net generating apparatus of Reactive Power Dispatch and method a few days ago.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, and provide the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, it not only saves computing time, and ensure that the optimality of scheduling scheme.

Technical solution of the present invention is: this generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, and this device comprises:

Scene optimization scheduling storehouse off-line forms module, and for carrying out cluster to wind power/photovoltaic generating historical data and be combined to form scene collection, the scheduling scheme calculating each scene is formed optimizes scene library;

Next day, state model matching module, for carrying out pattern matching according to the state 24 hours next day of prediction, formed scheduling scheme disaggregation next day;

Capacitor switching revision of option module, for revising capacitor switching scheme 24 hours next day;

Multiple attribute decision making (MADM) module, for obtaining optimal case and suboptimal design confession dispatcher reference.

Additionally provide a kind of generation method containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, the method comprises the following steps:

(1) cluster is carried out to the wind speed gathered, intensity of illumination and demand history data;

(2) wind speed good for cluster, intensity of illumination and payload historical data are carried out being combined to form scene collection;

(3) the idle model of exerting oneself of distributed power source is set up;

(4) for each scenario in scene collection, the genetic algorithm optimization Reactive Power Dispatch scheme of band elite retention strategy is adopted;

(5) optimal scheduling scheme and Run-time scenario will be tried to achieve merge and form scene optimization and dispatch storehouse;

(6) prediction wind-force 24 hours next day, photovoltaic are exerted oneself and payload;

(7) adopt Distance conformability degree computational methods to find out several group scenes similar with prediction case and corresponding Optimized Operation scheme thereof, form scheduling scheme disaggregation next day;

(8) fuzzy clustering is carried out by separating the capacitor 24 hours switching capacity concentrated; Revise the scheduling scheme in step (7);

(9) for each feasible solution obtained in step (8), propose multiple attribute decision making (MADM) index, utilize analytic hierarchy process (AHP) to determine each index weights, final calculating target function;

(10) optimal solution and suboptimal solution is drawn by sequence.

The present invention forms scene optimization scheduling storehouse by using Fuzzy clustering techniques off-line, adopt Distance conformability degree computational methods operational mode coupling, the several scenes the most close with actual conditions is found from scheduling storehouse, multiple attribute decision making (MADM) is utilized to find optimal scheduling scheme, thus not only save computing time, and ensure that the optimality of scheduling scheme.

Accompanying drawing explanation

Fig. 1 is the general frame of the present invention;

Fig. 2 is the flow chart of method of the present invention.

Embodiment

As shown in Figure 1, this generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, this device comprises:

Scene optimization scheduling storehouse off-line forms module, and for carrying out cluster to wind power/photovoltaic generating historical data and be combined to form scene collection, the scheduling scheme calculating each scene is formed optimizes scene library;

Next day, state model matching module, for carrying out pattern matching according to the state 24 hours next day of prediction, formed scheduling scheme disaggregation next day;

Capacitor switching revision of option module, for revising capacitor switching scheme 24 hours next day;

Multiple attribute decision making (MADM) module, for obtaining optimal case and suboptimal design confession dispatcher reference.

Preferably, described scene optimization scheduling storehouse off-line formation module also comprises:

Gather a large amount of historical datas of wind speed, intensity of illumination and load respectively, and consider load growth; Adopt Clustering Analysis Technology, respectively best fuzzy classified matrix and cluster centre are calculated to wind speed, intensity of illumination and payload historical data; Combining completing the wind speed of cluster, intensity of illumination and payload, setting up comprehensive Run-time scenario collection; To each set up Run-time scenario collection, solve optimal scheduling scheme; Run-time scenario and corresponding optimal scheduling scheme are merged, forms scene optimization scheduling storehouse.

Preferably, described next day, state model matching module also comprised:

Prediction 24 hours next day wind-force, photovoltaic, load are exerted oneself; Adopt Distance conformability degree computational methods to carry out pattern matching, form scheduling scheme disaggregation next day.

Preferably, described capacitor switching revision of option module also comprises:

According to the restriction of capacitor switching number of times, utilize Fuzzy clustering techniques, optimize capacitor switching scheme.

Preferably, described multiple attribute decision making (MADM) module also comprises:

Utilize analytic hierarchy process (AHP) determination decision index system weight; Optimal scheduling scheme is obtained to target function sequence.

Preferably, the plurality objective function of decision-making adopted, comprises loss, variation and gas turbine operation expense.

As shown in Figure 2, additionally provide a kind of generation method containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, the method comprises the following steps:

(1) cluster is carried out to the wind speed gathered, intensity of illumination and demand history data;

(2) wind speed good for cluster, intensity of illumination and payload historical data are carried out being combined to form scene collection;

(3) the idle model of exerting oneself of distributed power source is set up;

(4) for each scenario in scene collection, the genetic algorithm optimization Reactive Power Dispatch scheme of band elite retention strategy is adopted;

(5) optimal scheduling scheme and Run-time scenario will be tried to achieve merge and form scene optimization and dispatch storehouse;

(6) prediction wind-force 24 hours next day, photovoltaic are exerted oneself and payload;

(7) adopt Distance conformability degree computational methods to find out several group scenes similar with prediction case and corresponding Optimized Operation scheme thereof, form scheduling scheme disaggregation next day;

(8) fuzzy clustering is carried out by separating the capacitor 24 hours switching capacity concentrated; Revise the scheduling scheme in step (7);

(9) for each feasible solution obtained in step (8), propose multiple attribute decision making (MADM) index, utilize analytic hierarchy process (AHP) to determine each index weights, final calculating target function;

(10) optimal solution and suboptimal solution is drawn by sequence.

The present invention forms scene optimization scheduling storehouse by using Fuzzy clustering techniques off-line, adopt Distance conformability degree computational methods operational mode coupling, the several scenes the most close with actual conditions is found from scheduling storehouse, multiple attribute decision making (MADM) is utilized to find optimal scheduling scheme, thus not only save computing time, and ensure that the optimality of scheduling scheme.

For making the object of the embodiment of the present invention, technical scheme and advantage clearly, below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly described, obviously, described embodiment is the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.

As shown in Figure 1, first the embodiment of the present invention carries out cluster to the wind speed gathered, intensity of illumination and demand history data, calculates best fuzzy classified matrix and cluster centre.

Cluster analysis is a kind of effective ways solving phase recency between data, and main thought is the membership class carrying out defined variable by degree of membership, and the maximum membership degree according to each variable is classified.If X={x ₁, x ₂..., x ₂₄be the set of sample to be sorted, i.e. wind speed, intensity of illumination and payload historical data, n is classification samples number, and the target function of fuzzy clustering is:

$\min J(U,V)=\underset{k=1}{\overset{n}{\mathrm{\Σ}}}\underset{i=1}{\overset{c}{\mathrm{\Σ}}}{\left(u_{\mathrm{ik}}\right)}^m{\left|\right|x_k-v_i\left|\right|}^2$

In formula, c is cluster number; M is Weighted Index (usually getting m=2); v _iit is the i-th class cluster central value; u _ikfor x _kbelong to the subordinated-degree matrix of the i-th class; U={u _ikit is subordinated-degree matrix; d _ikfor variable x _karrive and cluster centre value v _ieuclidean distance; Target function J (U, V) is for all kinds of middle sample is to the Weighted distance quadratic sum of cluster centre.

Wind speed good for cluster, intensity of illumination and payload historical data are carried out being combined to form scene collection.Such as wind speed cluster number is a, and intensity of illumination cluster number is b, and payload cluster number is c, and the scene collection namely after combination has a*b*c scene number.

For each scenario in scene collection, adopt the genetic algorithm optimization Reactive Power Dispatch scheme of band elite retention strategy.Concrete steps are as follows.

1) distributed power source idle go out force modeling

The maximum of double-fed blower fan absorbing reactive power slip when depending on fan operation, active power export and maximum stator current.

$Q_{\mathrm{ab}\max }^{\mathrm{wind}}=\sqrt{{\left(\right|V\left|I_s^{\mathrm{wind}}\right)}^2-{\left(\frac{P_{\mathrm{wind}}}{1-s}\right)}^2}$

In formula: P _windfor active power; Slip when s is wind turbine steady operation; V is blower fan node voltage; for maximum stator current.The maximum of output reactive power determined by maximum rotor electric current and fan parameter.

$Q_{\mathrm{in}\max }^{\mathrm{wind}}=1-\frac{{\left|V\right|}^2(X_n^s+X_n^m)}{{|Z_n^s+Z_n^m|}^2}+\frac{\left|V\right|\left|Z_n^m\right|I_r^{\mathrm{wind}}\sin \mathrm{\γ}}{|Z_n^s+Z_n^m|}$

In formula: be respectively blower fan equivalent impedance and stator impedance; for maximum rotor electric current; γ is power-factor angle.Can be calculated as follows:

$\mathrm{\γ}={\cos }^{-1}\left(\frac{P_{\mathrm{wind}}{|Z_n^s+Z_n^m|}^2+{\left|V\right|}^2(R_n^s+R_n^m)}{(1-s)\left|V\right|\left|Z_n^m\right||Z_n^s+Z_n^m|I_r^{\mathrm{wind}}}\right)$

Photovoltaic cell can produce reactive power by the multiplex technique of inverter by during inverter access electrical network, and it is little to export impact to the active power of system.Photovoltaic generation exports, the maximum of absorbing reactive power is:

$Q_{\mathrm{in}\max }^{\mathrm{pv}},Q_{\mathrm{ab}\max }^{\mathrm{pv}}=\sqrt{{\left(\right|V\left|I_n^{\mathrm{pv}}\right)}^2-{\left(P_{\mathrm{pv}}\right)}^2}$

In formula: be respectively photovoltaic generation reactive power export maximum and absorb maximum; P _pvfor photovoltaic generation is gained merit injecting power, affect by natural resources; V is grid-connected node voltage; for inverter current.

Miniature gas turbine generally adopts magneto alternator to realize power delivery through rectifier rectification and grid-connected inverters, and adopts PQ uneoupled control, and therefore the idle Power generation limits of miniature gas turbine is mainly by inverter capacity-constrained, and reactive power auxiliary service value is:

$Q_{\mathrm{in}\max }^{\mathrm{MT}},Q_{\mathrm{ab}\max }^{\mathrm{MT}}=\sqrt{S^2-{\left(P_{\mathrm{MT}}\right)}^2}$

In formula: be respectively the idle output maximum of miniature gas turbine and absorb maximum; S is inverter capacity, P _mTfor miniature gas turbine meritorious go out force value.

2) based on the idle work optimization of genetic algorithm

To each set up Run-time scenario collection, take loss minimization as target function, idlely to exert oneself with wind power generation, photovoltaic generation is idle exerts oneself, gas turbine is idle exerts oneself and capacitor is idle exerts oneself as optimized variable, solve optimal scheduling scheme.Finally obtain the optimal scheduling scheme of each Run-time scenario in step 1.Optimization object function is:

F＝minP _loss

Constraints comprises: active power and reactive power trend Constraints of Equilibrium:

$\left\{\begin{array}{c}{P}_{\mathrm{Gi}}-P_{\mathrm{Li}}=V_i\underset{j=1}{\overset{\mathrm{Nb}}{\mathrm{\Σ}}}{V}_j(G_{\mathrm{ij}}\cos {\mathrm{\δ}}_{\mathrm{ij}}+B_{\mathrm{ij}}\cos {\mathrm{\δ}}_{\mathrm{ij}})\\ Q_{\mathrm{Gi}}+Q_{\mathrm{Ci}}-Q_{\mathrm{Li}}=V_i\underset{j=1}{\overset{\mathrm{Nb}}{\mathrm{\Σ}}}{V}_j(G_{\mathrm{ij}}\sin {\mathrm{\δ}}_{\mathrm{ij}}-B_{\mathrm{ij}}\cos {\mathrm{\δ}}_{\mathrm{ij}})\end{array}\right.$

Blower fan reactive power is exerted oneself restriction:

Q _WTmax≤Q _WT≤Q _WTmax

Photovoltaic system is idle exerts oneself restriction:

Q _PVmin≤Q _PV≤Q _PVmax

Miniature gas turbine is idle exerts oneself restriction:

Q _MTGmin≤Q _MTG≤Q _MTGmax

In genetic manipulation, adopt segment encoding mode, according to the kind determination segments of optimized variable, for the equipment of identical type, in section, carry out heritable variation operation, ensure that the population of stochastic generation farthest meets constraints.

After trying to achieve optimal scheduling scheme, merge with Run-time scenario and form scene optimization and dispatch storehouse.

When carrying out on-line optimizing scheduling, first prediction wind-force 24 hours next day, photovoltaic are exerted oneself and payload.Afterwards the running status per hour next day of prediction, dispatch the service conditions comparison in storehouse with scene optimization, adopt Distance conformability degree computational methods to find out several groups of similar scenes and the Optimized Operation scheme of correspondence thereof, form scheduling scheme disaggregation next day.Concrete grammar is as follows:

$F=\min \sqrt{{(x_1-x)}^2+{(y_1-y)}^2+{(z_1-z)}^2}$

Adopt Euclidean distance, use wind speed, intensity of illumination and payload three dimensions distance to find scene optimization to dispatch several groups of scenes the most similar to actual conditions in storehouse, form scheduling scheme disaggregation next day.

In order to meet the constraint of capacitor one day 24 hours switching frequency, carry out fuzzy clustering by separating the capacitor 24 hours switching capacity concentrated.Cluster centre quantity is that one day inner capacitor can switching maximum times.Substitute with this result the capacitor that next day, scheduling scheme solution was concentrated to exert oneself.The target function of fuzzy clustering is:

$\begin{array}{c}\min J(U,J)=\underset{k=1}{\overset{24}{\mathrm{\Σ}}}\underset{i=1}{\overset{c}{\mathrm{\Σ}}}{\left(u_{\mathrm{ik}}\right)}^m{\left(d_{\mathrm{ik}}\right)}^2\\ =\underset{k=1}{\overset{24}{\mathrm{\Σ}}}\underset{i=1}{\overset{c}{\mathrm{\Σ}}}{\left(u_{\mathrm{ik}}\right)}^2{\left|\right|x_k|-v_i|}^2\end{array}$

Propose multiple attribute decision making (MADM) index, calculated the weight of each index by analytic hierarchy process (AHP), to each the feasible solution calculating target function in obtained disaggregation.

Target function:

MinF=λ ₁p _loss+ λ ₂u _variation+ λ ₃c _{gas turbine}

In formula, three target functions are respectively network loss, variation size and gas turbine operation expense.Because three target function dimensions are different, standardization need be carried out.Standardization formula is as follows:

$x_i=\frac{x_i^{\′}-\min \left\{x_i^{\′}\right\}}{\max \left\{x_i^{\′}\right\}-\min \left\{x_i^{\′}\right\}}$

Analytic hierarchy process (AHP) step is as follows:

(1) first filled out a questionnaire for a certain evaluation object by expert

Wherein, b _ijreflect the significance level of index i relative to index j, adopt digital 1-9 and scale reciprocal thereof.

(2) judgment matrix is set up according to application form

$B=\left[\begin{array}{cccc}{b}_{11}& b_{12}& \·\·\·& b_{1m}\\ b_{21}& b_{22}& \·\·\·& b_{2m}\\ \·& \·& & \·\\ \·& \·& & \·\\ \·& \·& & \·\\ b_{m1}& b_{m2}& \·\·\·& b_{\mathrm{mm}}\end{array}\right]$

(3) the eigenvalue of maximum λ of judgment matrix B is asked _maxand characteristic of correspondence vector v

Bv＝λ _maxv

(4) each index weights is asked

$w_i=\frac{v_i}{\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{v}_k},i=\mathrm{1,2},\·\·\·,m$

(5) consistency check (m=1 does not need inspection when 2)

Random Consistency Ratio CR is

$\mathrm{CR}=\frac{\mathrm{CI}}{\mathrm{RI}}$

In formula: CI is coincident indicator; For Aver-age Random Consistency Index.

Coincident indicator CI can determine according to following formula

$\mathrm{CI}=\frac{{\mathrm{\λ}}_{\max }-m}{m-1}$

In formula: m is judgment matrix exponent number.

Aver-age Random Consistency Index, can look into following table and determine.

								0	1	2	3	4	5	6
																.58	.90	.12	.24	.32	.41	.45	.49	.52	.54	.55	.56	.57	.58

As CR<0.1, think that judgment matrix meets coherence request, otherwise just need to adjust the element value in judgment matrix, redefine weight.

After trying to achieve target function, sort by size, what come foremost is optimal solution, sorts in several solutions above as suboptimal solution, for dispatcher's reference.

The invention provides a kind of containing the wind power/photovoltaic generation distribution net generating apparatus of Reactive Power Dispatch and method a few days ago, it uses Fuzzy clustering techniques off-line to form scene optimization scheduling storehouse; Adopt Distance conformability degree computational methods operational mode coupling, from scheduling storehouse, find the several scenes the most close with actual conditions; Multiple attribute decision making (MADM) is utilized to find optimal scheduling scheme.It mainly contains the advantage of following several respects:

1, form scene optimization scheduling storehouse by a large amount of calculated off-line, ensure that the optimum of each scene dispatching result.Give and provide sufficient reference value in line computation.

2, a few days ago idle work optimization time by search in scene Optimized Operation storehouse, find fast and the scene that next day, operational mode was mated and scheduling scheme.Both save scheduling time, in turn ensure that the optimality of scheduling result.

3, by the cluster to capacitor switching scheme, capacitor switching number of times is limited.Greatly reduce the equipment loss because switching brings and labour cost.

4, by can obtain optimal scheduling scheme and suboptimum scheduling scheme, for dispatcher provides multiple reference to the sequence of target function simultaneously.

5, multiple attribute decision making (MADM) index meets the many index requirement of electrical network simultaneously, achieves comprehensive optimum.

The above; it is only preferred embodiment of the present invention; not any pro forma restriction is done to the present invention, every above embodiment is done according to technical spirit of the present invention any simple modification, equivalent variations and modification, all still belong to the protection range of technical solution of the present invention.

Claims (7)

1., containing the generating apparatus of wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, it is characterized in that: this device comprises:

Scene optimization scheduling storehouse off-line forms module, and for carrying out cluster to wind power/photovoltaic generating historical data and be combined to form scene collection, the scheduling scheme calculating each scene is formed optimizes scene library;

Next day, state model matching module, for carrying out pattern matching according to the state 24 hours next day of prediction, formed scheduling scheme disaggregation next day;

Capacitor switching revision of option module, for revising capacitor switching scheme 24 hours next day;

Multiple attribute decision making (MADM) module, for obtaining optimal case and suboptimal design confession dispatcher reference.

2. the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago according to claim 1, is characterized in that:

Described scene optimization scheduling storehouse off-line forms module and also comprises:

Gather a large amount of historical datas of wind speed, intensity of illumination and load respectively, and consider load growth; Adopt Clustering Analysis Technology, respectively best fuzzy classified matrix and cluster centre are calculated to wind speed, intensity of illumination and payload historical data; Combining completing the wind speed of cluster, intensity of illumination and payload, setting up comprehensive Run-time scenario collection; To each set up Run-time scenario collection, solve optimal scheduling scheme; Run-time scenario and corresponding optimal scheduling scheme are merged, forms scene optimization scheduling storehouse.

3. the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago according to claim 2, is characterized in that:

Described next day, state model matching module also comprised:

Prediction 24 hours next day wind-force, photovoltaic, load are exerted oneself; Adopt Distance conformability degree computational methods to carry out pattern matching, form scheduling scheme disaggregation next day.

4. the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago according to claim 3, is characterized in that:

Described capacitor switching revision of option module also comprises:

According to the restriction of capacitor switching number of times, utilize Fuzzy clustering techniques, optimize capacitor switching scheme.

5. the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago according to claim 4, is characterized in that:

Described multiple attribute decision making (MADM) module also comprises:

Utilize analytic hierarchy process (AHP) determination decision index system weight; Optimal scheduling scheme is obtained to target function sequence.

6. the generating apparatus containing wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago according to claim 5, is characterized in that:

The plurality objective function of decision-making adopted, comprises loss, variation and gas turbine operation expense.

7., containing the generation method of wind power/photovoltaic generation distribution net Reactive Power Dispatch a few days ago, it is characterized in that: the method comprises the following steps:

(1) cluster is carried out to the wind speed gathered, intensity of illumination and demand history data;

(2) wind speed good for cluster, intensity of illumination and payload historical data are carried out being combined to form scene collection;

(3) the idle model of exerting oneself of distributed power source is set up;

(4) for each scenario in scene collection, the genetic algorithm optimization Reactive Power Dispatch scheme of band elite retention strategy is adopted;

(5) optimal scheduling scheme and Run-time scenario will be tried to achieve merge and form scene optimization and dispatch storehouse;

(6) prediction wind-force 24 hours next day, photovoltaic are exerted oneself and payload;

(7) adopt Distance conformability degree computational methods to find out several group scenes similar with prediction case and corresponding Optimized Operation scheme thereof, form scheduling scheme disaggregation next day;

(8) fuzzy clustering is carried out by separating the capacitor 24 hours switching capacity concentrated; Revise the scheduling scheme in step (7);

(9) for each feasible solution obtained in step (8), propose multiple attribute decision making (MADM) index, utilize analytic hierarchy process (AHP) to determine each index weights, final calculating target function;

(10) optimal solution and suboptimal solution is drawn by sequence.