CN105552941A - Distributed power supply peak regulation capacity optimization method - Google Patents

Abstract

The invention provides a distributed power supply peak regulation capacity optimization method. An annual load duration curve is drawn according to an annual load curve, and a distribution network is planned under the next peak regulation load level of the maximum load. Therefore, the distributed power supply peak regulation capacity optimization method provided by the invention is used for reducing the investment funds and improving the equipment utilization rate.

Description

A kind of distributed power source peak optimization method

Technical field

The present invention proposes a kind of distributed power source peak optimization method, belongs to power distribution network Peak Shaving.

Background technology

According to The National Electric Power Communication Center's statistics, China's each department peak load of grid growth reaches more than 10, and low ebb load amplification is but no more than 5 percent, even there is low ebb load negative growth phenomenon.Carrying out load peak valley according to the minimum daily load rate in somewhere analyzes known, and summer, average minimum daily load rate was 0.561, and winter is 0.59, and peak-valley difference reaches 25GW.In other words, load peak-valley difference is increasing, peak regulation arduous task, and in the urgent need to considering the control of peak load in Electric Power Network Planning, this is conducive to the benign development of electrical network.

Traditional distribution network planning problem refers to and is meeting under the prerequisite retrained customer power supply and the network operation, seek the decision variable (path and size etc. of transformer station position and capacity, feeder line) of one group of optimum, make investment, operation, maintenance, network loss and reliability failure costs sum minimum.Be plan based on peak load as conventional electrical distribution network planning is standardized, planning investment is large like this, very flexible, and utilization rate of equipment and installations after planning is low.

Due to distributed power source have economy, flexibly, environmental protection and delay the advantages such as distribution network construction, more and more applied.Have in prior art according to year based model for load duration curve, based on branch road overload situations under different peak load level, adopt a kind of heuritic approach to carry out the addressing of distributed power source, wherein consider the peak regulation effect of oil-burning machine and gas engine respectively.Have consider electric automobile access network time, electric automobile charging station containing V2G function can utilize idle electric automobile to be wired back by back electrical energy net in peak times of power consumption as energy storage device, realize electric automobile and participate in power distribution network peak regulation as movable type distribution energy-storage units, achieve the optimization of power distribution network.

The Changing Pattern of load curve is the basis of distribution network planning.The load character of certain city newly-established high speed collar region is that office office building is more, is secondly residential quarter, does not have factory, analyze below in conjunction with concrete data from curve tendency, peak-valley ratio and duration of peaking time to part throttle characteristics.Shown in formula specific as follows:

$\%P=\frac{P_{max}-P_{min}}{P_{max}}\×100\%---\left(1\right)$

ΔT＝T _90％(2)

In formula: %P is peak-valley ratio, P _maxfor peak load, P _minfor minimum load, Δ T is duration of peaking time, T _90%be 90% peak load duration, be defined as duration of peaking time.

For the feature of the peak feature and peak of finding out different time, the typical daily load of different times in a year is selected to analyze, as shown in Figure 1.

As seen from Figure 1, except daily load curve in July 20 has except single-peak response, all there is morning peak and evening peak in four days in all the other; The situation that in bimodal load, morning peak is higher than evening peak is more; But along with temperature reduce, December 20 evening peak load higher than morning peak; The peak-valley ratio that different load curve peak features is corresponding differs greatly, the peak-valley ratio on July 20 with single-peak response reaches 78%, and evening peak is 62% higher than the peak-valley ratio on December 20 of morning peak, three days peak-valley ratios that morning peak is higher than evening peak are respectively 52.5%, 64% and 74%, therefore, different peaks is needed.

Summary of the invention

Goal of the invention: the present invention proposes a kind of distributed power source peak optimization method, utilizes the distributed power source in power distribution network to exert oneself and carries out peak regulation, improves flexibility and the capacity utilization of peak regulation.

Technical scheme: the present invention proposes a kind of distributed power source peak optimization method, comprises the following steps:

1) distributed area and the probability of peak-valley ratio is counted according to historical data, and the relation between the cost of foundation generating, transmission of electricity and capacity;

2) variance and duration of peaking time are normalized, draw target function expression formula;

3) in each peak-valley ratio interval in this interval selected the load curve of peak load point place that day as initial load level;

4) carry out chromosome coding, produce initial population;

5) calculate the calculating of fitness, and carry out the comparison of fitness, retain larger fitness value;

6) judge whether to meet end condition, if meet, jump procedure 7); Otherwise start selection, intersect, make a variation and jump to step 5);

7) draw optimum population, compare in colony, two schemes best with economic benefit that output variance is minimum respectively.

Preferably, described step 2) in the target function that obtains after normalized be:

f ₂＝min(C _c，C _j)

$C_c=(k_f+k_s)\·\mathrm{\Δ}P+f(\mathrm{\Δ}P,T)+\underset{j=1}{\overset{n}{\Σ}}{\mathrm{kP}}_{DGj}$

$g_i=a_i{P}_i^2+b_i{P}_i+c_i$

$C_j=\underset{i=1}{\overset{n}{\Σ}}{k}_{jj}{g}_i$

In formula, f ₂for the target function after normalization, (k _f+ k _s) be unit capacity cost of electricity-generating, Transmission Cost cost coefficient, Δ P is optimizing capacity, and f (Δ P, T) is interruption cost, relevant with optimizing capacity and duration, for distributed power source investment cost; g _ifor the amount of carbon dioxide (ton) reduced discharging, a _i, b _i, c _ifor function coefficients, can rule of thumb draw; p _ifor distributed power source optimizing capacity; k _jjfor the economic profit coefficients that unit carbon dioxide emission reduction amount brings.

Preferably, described step 2) also comprise following constraints:

L _min≤L≤L _max

In formula, L is load level, L _mina year minimum load, L _maxit is annual peak load;

P _G≤P _max

P in formula _gdistributed power source reserve capacity, P _maxit is Peak Load;

％P _imin≤％P _i≤％P _imax

%P in formula _ifor peak-valley ratio, %P _imaxmaximum peak-valley ratio, %P _iminit is minimum peak-valley ratio;

ΔT _imin≤ΔT _i≤ΔT _imax

Δ T in formula _ifor duration of peaking time, Δ T _imaxfor peak-peak load duration, Δ T _iminfor minimum peak load duration.

Preferably, described step 4) middle chromosome coding employing real coding, in units of 0.1MW, chromosome length equals power distribution network zoning number.

Preferably, described step 6) in end condition be that iterations reaches maximum iteration time.Described step 6) described in be chosen as optimum maintaining strategy.Described step 6) described in intersect for single-point intersection.

Beneficial effect: the present invention draws out a year based model for load duration curve according to yearly load curve, carries out distribution network planning under a peak value regulates load level at maximum load.Therefore present invention reduces investment amount, and improve utilization rate of equipment and installations.

Accompanying drawing explanation

Fig. 1 is certain circle typical day load curve figure at a high speed;

Fig. 2 is photovoltaic power curve figure;

Fig. 3 is wind-driven generator power curve figure;

Fig. 4 is the equivalent load curve chart after access solar energy power generating;

Fig. 5 is the equivalent load curve chart after access wind-driven generator;

Fig. 6 is access different capabilities solar photovoltaic power equivalent load figure;

Tu7Wei Mou city high speed collar region divides figure;

Fig. 8 is operational flowchart of the present invention;

Fig. 9 is the equivalent load curve chart after optimizing capacity configuration.

Embodiment

Below in conjunction with the drawings and specific embodiments, illustrate the present invention further, these embodiments should be understood only be not used in for illustration of the present invention and limit the scope of the invention, after having read the present invention, the amendment of those skilled in the art to various equivalents of the present invention has all fallen within the application's claims limited range.

Solar energy and wind energy belong to clean reproducible energy, solar energy power generating and wind power generation are all being greatly developed in the whole world, produce larger impact in distributed power generation field by the equivalent load of power distribution network, first the power producing characteristics of solar energy power generating and wind power generation is analyzed below.

The typical power curve of photovoltaic cell as shown in Figure 2, the at a time actual P that exerts oneself of t _tcan be expressed as:

$P_t=P_{stc}\frac{I_{rt}}{I_{stc}}\[1+{\mathrm{\α}}_T(T_t-T_{stc})\]---\left(3\right)$

In formula: T _stcthe exerting oneself of photovoltaic panel under reference condition, corresponding intensity of solar radiation I _stcfor 1000W/m2, temperature T _stc=25 DEG C; I _rtfor the intensity of solar radiation of t reality, T _tfor the temperature of t photovoltaic panel.The actual main impact being subject to intensity of solar radiation and temperature of exerting oneself of photovoltaic cell can be found out.

As seen from Figure 2 at noon ten two time temperature the highest, intensity of illumination is maximum, and photovoltaic generating system is exerted oneself larger; Sooner or later intensity of illumination is very low, exerts oneself also little.

The wind energy power that wind turbine sends can be expressed as:

$W_{\left(v\right)}=1/2{\mathrm{\ρAV}}_r^3{C}_p{(V/V_r)}^3---\left(4\right)$

In formula: W _(v)for blower fan goes out force level, P _rfor blower fan rated power, ρ is atmospheric density, and A is wind wheel sweeping area, and V is wind speed, V _rfor rated wind speed, C _pfor the power coefficient of wind wheel.

Fig. 3 is certain wind-driven generator power curve on July 20th, 2013, and can find out that the minimum windspeed that this wind-driven generator is exerted oneself is 6m/s, the wind speed of whole day does not all reach rated wind speed, so wind power generation output is the curve continuing change.Due to daytime especially noon, wind speed is lower, and causing generating now to be exerted oneself is zero substantially; Because wind speed is comparatively large time at dusk, wind-driven generator is exerted oneself larger.

In Fig. 1 on July 10th, 2013 load based on; capacity of choosing respectively is 500kW photo-voltaic power supply and wind-driven generator; maximum output calculates by 80%; the threshold wind velocity of wind-driven generator is 6m/s; wind speed under rated power is 15m/s; shutdown wind speed is 25m/s, and respectively as shown in Figure 4 and Figure 5, peak regulation Contrast on effect result is as shown in table 1 for the curve before and after equivalent load.

Table 1: peak regulation effectiveness comparison

Can find in conjunction with above chart, solar energy power generating power curve is close with this region load curve, and peak load significantly reduces, although minimum load has reduction to a certain degree, but peak-valley difference and peak-valley ratio reduce greatly, and duration of peaking time increases greatly, peak regulation Be very effective.The change of wind power generation output curve is then contrary with load, has anti-Peak Load Adjustment.So, solar energy power generating can be adopted to carry out peak regulation.

From analysis before, the peak-valley ratio that different load peak characters is corresponding and duration of peaking time have difference, for the Peak Load Adjustment playing distributed power source needs to be optimized its capacity, and need when optimizing to consider different peak characters.

The Peak Load Adjustment of distributed power source mainly utilizes its power producing characteristics and part throttle characteristics to carry out comprehensively cutting down the peak load value of equivalence, and therefore effective Peak Load Adjustment is mainly reflected on the change indicator of peak-valley ratio and duration of peaking time.Along with the difference of access distributed power source capacity, it is also different that the peak-valley ratio brought reduces degree, for actual load on July 10th, 2013, in the distributed power source situation of access different capabilities, distributed power source exert oneself comprehensive with load after equivalent load curve as shown in Figure 6.Can find out, it is more that distributed power source capacity increase along with access can make peak-valley ratio reduce, but when distributed power source capacity increases to a certain degree, peak-valley ratio will remain unchanged, then along with the capacity of access distributed power source continues to increase, peak-valley ratio starts again to increase.Meanwhile, also can increase along with investment because distributed power source capacity increases.Therefore, the capacity considering how choose reasonable distributed power source is needed.

Generally, duration of peaking time considers according to the duration of more than 90% load of peak load, peak load 90% also referred to as reference load.Can find out, when installing photovoltaic generating system according to 0.6MW, the duration of peaking time of equivalent load became 7 hours from original 4 hours, and load curve becomes more steady, and duration of peaking time is longer, and peak regulation effect is more obvious.

The present invention, when being optimized configuration capacity and choosing, first counts distributed area and the probability of peak-valley ratio according to history daily load curve; The relation of generating, Transmission Cost and capacity is set up by electric company statistics.

Then equivalent load curve variance C is set up _f, duration of peaking time rate C _fh, distributed power source cost C _cthe energy-saving benefit C accessed with distributed power source _jthe Model for Multi-Objective Optimization of four indexs.At this, we carry out a definition: we calculate with the load in one day 24 integral point moment when carrying out peak-valley ratio and duration of peaking time computational analysis, and we claim the load in integral point moment to be comparison load.

First target function is set up:

f ₁＝min(C _f)(5)

$C_f=\underset{i=1}{\overset{2}{\Σ}}4{(P_i-P_{av})}^2---\left(6\right)$

C _fh＝ΔT/T

In formula, f ₁for target function one, P _ifor the comparison load in each moment, P _avfor the mean value of each comparison load, Δ T is duration of peaking time.

Duration of peaking time we can define like this: using 90% level of peak load as with reference to load, in one day, the load value in 24 integral point moment is as comparison load.When the comparison load in adjacent two moment is all large than reference load time, then think that load within this hour is all on reference load, then the duration increases by one hour; When the comparison load one in adjacent two moment is larger than reference load, one when being not more than reference load, then the duration increases half an hour; If the comparison load in two moment is all not more than reference load, then the duration does not increase.When calculating, conveniently calculate us and be normalized, concrete grammar is as follows:

When variance and duration of peaking time rate satisfy condition, reach can be excellent object time, we consider cost and energy-saving benefit again:

f ₂＝min(C _c，C _j)(7)

$C_c=(k_f+k_s)\·\mathrm{\Δ}P+f(\mathrm{\Δ}P,T)+\underset{j=1}{\overset{n}{\Σ}}{\mathrm{kP}}_{DGj}---\left(8\right)$

$g_i=a_i{P}_i^2+b_i{P}_i+c_i$

$C_j=\underset{i=1}{\overset{n}{\Σ}}{k}_{jj}{g}_i$

In formula, f ₂for target function two, (k _f+ k _s) be unit capacity cost of electricity-generating, Transmission Cost cost coefficient, Δ P is optimizing capacity, and f (Δ P, T) is interruption cost, relevant with optimizing capacity and duration, for distributed power source investment cost; g _ifor the amount of carbon dioxide (ton) reduced discharging, a _i, b _i, c _ifor function coefficients, can rule of thumb draw; p _ifor distributed power source optimizing capacity; k _jjfor the economic profit coefficients that unit carbon dioxide emission reduction amount brings.

Cost of the present invention we consider that cost of electricity-generating, Transmission Cost, interruption cost and distributed power source investment cost are carried out peak value and regulated the choosing of load.

(1) cost of electricity-generating and Transmission Cost

When often using kilowatt-hour less in power distribution network, then can delay electricity generation system, construction cost that transmission system is certain.Due to generating expense and transmission charges in current electric company by certain standard, so we can directly use this standard, be set to k respectively _fand k _scalculate, unit ten thousand yuan/kWh.Cost is directly relevant to capacity, thus can obtain:

C _f+s＝(k _f+k _s)·ΔP

In formula, C _f+sfor generating and Transmission Cost.

(2) interruption cost

So-called loss of outage refers to the loss caused national economy because distribution system has a power failure.Comprising the economic loss that the Custom interruption cost caused user and power department self cause because of power failure.Because the diversity of power consumer, loss of outage assessment is a complicated job, this is done to a simplification herein, only considers its economy, as long as namely think that load level exceedes planned load, then can cause loss of outage.

Loss of outage is main relevant with the duration of capacity and this kind of capacity level.By chapter 2 content we can draw a year based model for load duration curve, according to this curve, as long as determine a capacity level, just can draw a duration, again according to the loss of outage of unit capacity under this kind of capacity, just the interruption cost required when this load level carries out distribution network planning can be calculated.

Consider that resident's loss of outage calculates herein, loss of outage influencing factor is a lot, relevant with user characteristics, power-off characteristic, user type and the degree of dependence of user to electric energy.User investigation method is the best-evaluated method of carrying out loss of outage calculating based on user data, because questionnaire is estimated to contain above many Considerations by inquiry, thus provides the data basis that we calculate loss of outage.The loss of outage obtained with inquiry agency is benchmark, uses regression algorithm to calculate, draw to increase at load, reduce under interruption cost.

Loss of outage claimable amount ρ (unit/(kWh)) is defined as:

$\mathrm{\ρ}=\frac{C}{D}\×T$

In formula, C is the power failure interruption cost of 1 hour; D is year power consumption (unit kWh); T is year electricity consumption hourage (unit h).

Assumed load variation percentage be X, loss of outage is Y (unit: unit/kWh), and load change percentage and loss of outage are all considered as stochastic variable, by X and Y correlation coefficient r computing formula we can obtain:

$r=\frac{n\ΣXY-\ΣX\ΣY}{\sqrt{n\ΣX^2-{(\ΣX)}^2}\sqrt{n\ΣY^2-{(\ΣY)}^2}}$

By result of calculation, we can draw the relation between loss of outage and load change, thus determine the model between load change and loss of outage.

(3) distributed power source expense

Introducing distributed power source carries out peak regulation and must introduce distributed power source expense.The investment of distributed power source and the capacity of distributed power source closely related, this chapter content, due to the uncertainty of optimizing capacity, so do a simplification herein, supposes the investment cost C of distributed power source _dGwith distributed power source capacity P _dGjlinear, wherein k is coefficient:

$f\left(C_{DG}\right)=\underset{j=1}{\overset{n}{\Σ}}{\mathrm{kP}}_{DGj}$

Next constraints is provided.Because target function starts to calculate based on peak load level, and stop when the value of load level lower than some settings, it must meet:

L _min≤L≤L _max

In formula, L is load level, L _mina year minimum load, L _maxit is annual peak load.

P _G≤P _max

P in formula _gdistributed power source reserve capacity, P _maxit is Peak Load.

％P _imin≤％P _i≤％P _imax

%P in formula _ifor peak-valley ratio, %P _imaxmaximum peak-valley ratio, %P _iminit is minimum peak-valley ratio.

ΔT _imin≤ΔT _i≤ΔT _imax

Δ T in formula _ifor duration of peaking time, Δ T _imaxfor peak-peak load duration, Δ T _iminfor minimum peak load duration.

Due to the load curve in an interval peak-valley ratio relatively, its load curve is also more similar, so the load curve that we can choose the internal loading in interval maximum that day represents the load level within this interval, and as initial load.

Then chromosome coding is carried out.Coding is the method actual solution of problem and the chromosome of genetic algorithm connected.Conventional coded system has binary coding, symbolic coding and floating-point encoding.In order to simple and convenient, we adopt real coding.

Table 2: real coding chromosome

2

5

...

...

3

1

We are in units of 0.1MW, then first 2 represent 0.2MW, and 5 in form represents 0.5MW, and the rest may be inferred.Chromosome length we the number in region that just can divide according to certain city specify: if divide 10 regions into, then chromosome length is just 10, is divided into 15 regions, then chromosome length is just 15.Carry out coding so simple and convenient, the speed of service also than comparatively fast, can find out the size of connect capacity intuitively.

Genetic algorithm obtains next step search information by solving target function value, and the use of target function value realizes by calculating fitness function value size.Concrete operations are first decoded by chromosome, then calculates the corresponding individual target function value of this chromosome, then obtain fitness by target function value by certain transformation rule.

Because target function is a positive number certainly, and be 01 after being normalized, due to required be minimum value, then can use 1 with the difference of target function as fitness function.F (x)=1-f (x), asks f (x) minimum value, asks F (x) maximum exactly.

Draw distributed power source capacity after decoding, now in conjunction with the load curve of peak day, consider the equivalent load curve after access solar photovoltaic power, judge whether constraints now meets.

Natural evolutionary process follows the principle of " survival of the fittest ", and namely high to adaptive capacity to environment species will survive, and procreation is of future generation; And the low species of adaptive capacity to be genetic to follow-on possibility just little, be slowly eliminated.The process implementation of this survival of the fittest of selection opertor simulation in genetic algorithm.

The selection opertor that the present invention adopts adopts optimum maintaining strategy, namely when carrying out selection operation, we consider that the individuality that filial generation fitness function is the highest directly enters the next generation, and what also namely ensure every generation has individuality can enter the next generation most to proceed genetic manipulation.

Interlace operation in genetic algorithm refers to be carried out according to certain rules exchanging by the portion gene on the chromosome of two pairings and obtains two new individualities.Interlace operation plays an important role in genetic algorithm, is to produce new individual main method.Crossover operator herein adopts single-point to intersect, and first carries out random pair between two to the individuality in colony.The individuality good to every a pair pairing, stochastic generation crosspoint, then according to the portion gene on crossover probability chiasmatypy.

Mutation operation in genetic algorithm replaces chromosomal certain or some other allele of gene, and then obtain new individuality, and the mutation operation that the present invention carries out is as follows:

To each gene position of individuality, specify change point according to mutation probability; Separate according to concrete numbering generation field corresponding on current chromosome change point, determine candidate solution.Suppose initial solution and change point situation as shown in table 3:

Table 3: mutation operation

0

1

6

3

5

Suppose that there are 6 regions in somewhere, gene position value then on chromosome is 0-6, the distributed power source of 0.5MW capacity will be accessed in somewhere, be that a unit calculates with 0.1MW, then we can show that chromosome length is 5, as shown shown in 3-7, hypothetical gene value be 6 gene position be change point, the field that then can produce is separated, disaggregation is (01035) (01135) (01235) (01335) (01435) (01535), then selects neighborhood solution to concentrate optimum individuality.

Search end condition reaches maximum iteration time when iterations.Whole flow process as shown in Figure 8.

Next how set forth the present invention with an example utilizes genetic algorithm to carry out the optimization of distributed power source capacity.In this Section is in conjunction with certain city's distribution network planning content, choose interior 10 sub regions divided of high speed collar region in certain city as research object, industrial cousumer is not had at a high speed in circle, suppose the interior 10 sub regions uniform illumination of circle at a high speed, temperature is the same, and the situation choosing morning peak in peak character higher than evening peak is analyzed.By gathering calculating, can find to enclose at a high speed have within 266 days, have this peak character in interior 1 year, the load curve with such peak character being divided according to peak-valley ratio, and is divided in five intervals, specifically as shown in table 4.Due to the load curve in an interval peak-valley ratio relatively, its load curve is also more similar, so the load curve that we can choose the internal loading in interval maximum that day represents the load level within this interval.

Table 4: the probability distribution of peak-valley ratio

Peak-valley ratio	0-20％	20％-40％	40％-60％	60％-80％	80％-100％
						Probability	0％	10.15％	55.64％	22.94％	11.27％

Loss of outage data in our list of references [48] of data of loss of outage reparation model, carry out the calculating of loss of outage model.

Table 5: typical user's loss of outage statistical form

Load change percentage (compared with maximum load)	Resident's loss of outage claimable amount (unit/kWh)
		0	6.4，9，7.1
-10％	7.1，8.9，10
		-20％	9，9.5，10.7
-30％	11，12.3，14

Computing formula according to aforementioned correlation coefficient r can calculate resident's correlation coefficient r _g=0.742.Inquiry coefficient correlation form can know to there is certain linear relationship between loss of outage and load change percentage, so we can using load change as independent variable, loss of outage is independent variable, sets up linear regression model (LRM):

Y＝a+bX

The parameter can obtaining resident according to table 5 is respectively: a _j=1.155b _j=1.744;

The relational expression can obtained again between resident's loss of outage and Load Regulation is:

$Y_j=1.155+1.744X_g(-0.3\stackrel{\·}{E}{X}_g\stackrel{\·}{E})$

Because load and duration specify according to year based model for load duration curve, when calculating, we can directly draw from curve.When carrying out energy-saving benefit calculating, we consider to use the model of power and carbon emission in document [49] to calculate.

Initial parameter arranges as follows: Population Size is set to 100, iterations N we be chosen for 1000, if crossover probability is too large, result just has certain randomness, if too little, restrain slower, we choose 0.45 herein, and mutation probability is generally between 0.01 to 0.1, and we elect 0.06 as herein.Increase load newly according to certain electric company 12 period 110kV and following specific investment cost to calculate according to 0.003kW/ unit, then can reduce 110kV and with the power grid construction investment cost 2.51 hundred million yuan issuing transmission facility, that is we can regard an entirety as sending out transmission of electricity, then the electricity of 1kW often saved by power distribution network, k _f+ k _s=0.033 (ten thousand yuan/kW), when carrying out peak-valley ratio calculating, the equivalent load obtained after considering access solar photovoltaic power, a bodge is 0.1MW.

The situation of demand fulfillment peak load time owing to carrying out distribution network planning, in conjunction with yearly load curve, first we show that annual peak load is in this sky on August 20th, 2013, the peak-valley ratio of this day is 59%, drop within 40%60% this interval, so the load curve choosing this day represents all load curves within this interval.When genetic algorithm stops time, we can show that the distributed power source Peak Load in 10 regions is respectively:

Table 6: distributed power source Peak Load distribution table

Numbering	A	B	C	D	E	F	G	H	I	J
											Optimum by variance	3	2	0	2	1	2	10	1	1	4
By economic optimum	2	3	1	1	3	1	8	2	1	3

Because the probability occurred between peak-valley ratio 020% is 0, so we do not consider this interval situation.Thus, we can show that the distributed power source that different peak-valley ratio draws distributes capacity rationally respectively:

Table 7: different peak-valley ratio optimizing capacity configuration (optimum by variance)

Numbering	A	B	C	D	E	F	G	H	I	J
											20％-40％	1	2	1	2	0	3	8	2	1	1
40％-60％	2	2	0	2	1	2	9	1	1	3
											60％-80％	1	3	2	4	2	1	7	2	3	1
80％-100％	2	1	1	2	4	3	8	3	1	2

So in conjunction with the probability between each peak-valley ratio location, we according to table 7 can draw according to variance optimum draw distribute capacity rationally, as shown in table 8.

Table 8: final optimization pass configuration capacity (optimum by variance)

Numbering	A	B	C	D	E	F	G	H	I	J
											Capacity	2	2	1	2	1	2	8	2	2	3

Table 9: different peak-valley ratio optimizing capacity configuration (economy is more excellent)

Numbering	A	B	C	D	E	F	G	H	I	J
											20％-40％	0	1	2	2	1	2	8	3	2	1
40％-60％	1	2	0	2	1	1	9	2	1	2
											60％-80％	2	4	3	2	2	2	7	2	2	1
80％-100％	2	2	3	1	2	3	8	3	1	3

In like manner we can draw in the scheme meeting the economy optimum that variance preferably basis draws, as shown in Table 9 and Table 10.

By result of calculation, we are for the load condition of peak day, can show that distributed photovoltaic power is exerted oneself the equivalent load situation of situation and load, as shown in Figure 9.

As can be seen from figure we, access solar photovoltaic power after, be 38.7% by the more excellent peak-valley ratio drawn of variance, duration of peaking time is 6.2 hours, and economic benefit is 223.2 ten thousand yuan; Be 40% according to the more excellent peak-valley ratio that obtains of economy, duration of peaking time is 5.9 hours; Economic benefit is 231.9 ten thousand yuan.Capacity configuration is optimized when comprehensively we choose variance optimum.Optimized by distributed power source capacity known when peak load level 81%, when namely the duration is 700h, now can obtain comprehensively preferably distributing effect rationally.

Table 10: final optimization pass configuration capacity (economy is more excellent)

Numbering	A	B	C	D	E	F	G	H	I	J
											Capacity	1	2	1	3	1	2	8	2	1	2

Table 11: optimum results compares

	Optimum by peak regulation effect	Optimum by economy
			Capacity (MW)	2.5	2.3
Generating and Transmission Cost (ten thousand yuan)	657.2	608.1
			Interruption cost (ten thousand yuan)	201.4	135.2
Distributed power source investment cost (ten thousand yuan)	300	276
			Energy-saving benefit (ten thousand yuan)	66.4	35.0
Total economic benefit (ten thousand yuan)	223.2	231.9

Claims (7)

1. a distributed power source peak optimization method, is characterized in that, comprises the following steps:

1) distributed area and the probability of peak-valley ratio is counted according to historical data, and the relation between the cost of foundation generating, transmission of electricity and capacity;

2) variance and duration of peaking time are normalized, draw target function expression formula;

3) in each peak-valley ratio interval in this interval selected the load curve of peak load point place that day as initial load level;

4) carry out chromosome coding, produce initial population;

5) calculate the calculating of fitness, and carry out the comparison of fitness, retain larger fitness value;

6) judge whether to meet end condition, if meet, jump procedure 7); Otherwise start selection, intersect, make a variation and jump to step 5);

7) draw optimum population, compare in colony, two schemes best with economic benefit that output variance is minimum respectively.

2. distributed power source peak optimization method according to claim 1, is characterized in that, described step 2) in the target function that obtains after normalized be:

f ₂＝min(C _c，C _j)

$C_c=(k_f+k_s)\·\mathrm{\Δ}P+f(\mathrm{\Δ}P,T)+\underset{j=1}{\overset{n}{\Σ}}{\mathrm{kP}}_{DGj}$

$g_i=a_i{P}_i^2+b_i{P}_i+c_i$

$C_j=\underset{i=1}{\overset{n}{\Σ}}{k}_{jj}{g}_i$

In formula, f ₂for the target function after normalization, (k _f+ k _s) be unit capacity cost of electricity-generating, Transmission Cost cost coefficient, Δ P is optimizing capacity, and f (Δ P, T) is interruption cost, relevant with optimizing capacity and duration, for distributed power source investment cost; g _ifor the amount of carbon dioxide (ton) reduced discharging, a _i, b _i, c _ifor function coefficients, can rule of thumb draw; p _ifor distributed power source optimizing capacity; k _jjfor the economic profit coefficients that unit carbon dioxide emission reduction amount brings.

3. distributed power source peak optimization method according to claim 2, is characterized in that, described step 2) also comprise following constraints:

L _min≤L≤L _max

In formula, L is load level, L _mina year minimum load, L _maxit is annual peak load;

P _G≤P _max

P in formula _gdistributed power source reserve capacity, P _maxit is Peak Load;

％P _imin≤％P _i≤％P _imax

%P in formula _ifor peak-valley ratio, %P _imaxmaximum peak-valley ratio, %P _iminit is minimum peak-valley ratio;

ΔT _imin≤ΔT _i≤ΔT _imax

Δ T in formula _ifor duration of peaking time, Δ T _imaxfor peak-peak load duration, Δ T _iminfor minimum peak load duration.

4. distributed power source peak optimization method according to claim 1, is characterized in that, described step 4) middle chromosome coding employing real coding, in units of 0.1MW, chromosome length equals power distribution network zoning number.

5. distributed power source peak optimization method according to claim 1, is characterized in that, described step 6) in end condition be that iterations reaches maximum iteration time.

6. distributed power source peak optimization method according to claim 1, is characterized in that, described step 6) described in be chosen as optimum maintaining strategy.

7. distributed power source peak optimization method according to claim 1, is characterized in that, described step 6) described in intersect for single-point intersection.