CN104091041A - High-order-moment based generated power estimation method and system - Google Patents

Abstract

The invention provides a high-order-moment based generated power estimation method and system. The high-order-moment based generated power estimation method comprises the steps as follows: a plurality of sample groups are generated, one sample group which has the smallest characteristic deviation value f with original time series data is determined, wind speed data is simulated according to a wind speed Weibull probability distribution function corresponding to the determined sample group, and generated power is estimated according to the simulated wind speed data. According to the provided generated power estimation method, the accuracy degree for power estimation of wind power generation or photovoltaic power generation can be increased.

Description

Generated output evaluation method and system based on High Order Moment

Technical field

The present invention relates to energy technology field, relate in particular to a kind of generated output evaluation method and system based on High Order Moment.

Background technology

Along with day by day increasing the weight of of the problem such as lack of energy and environmental pollution, distributed power generation taking the regenerative resource such as wind, light as primary energy has become research focus and the cutting edge technology of current power engineering and energy field, and wind-force, extensive grid-connected planning and design of power system, the economic security of making of photovoltaic distributed generating are moved and faced a severe challenge.The natural resources such as wind, light has timing, undulatory property and randomness, makes distributed power source generated output present intermittent feature.In order to further investigate wind-force, the impact of photovoltaic distributed power grid on electric system, in controlling, Power System Planning, scheduling and operation take into full account the intermittence of wind-force, photovoltaic distributed power supply, formulation the exert oneself effective means of random fluctuation of a large amount of distributed power sources of dissolving, must set up the mathematical model that can reflect comparatively accurately wind-force, photovoltaic generation power probability statistics feature, to realize the real simulation to wind-force, photovoltaic generation.

Beirut American university professor S.H.Karaki etc. has announced a kind of multimode modelling method of probabilistic.The method by wind-power electricity generation exert oneself, photovoltaic generation is exerted oneself etc., and multiple discrete Qualitative state really that continuous nondeterministic statement changes into according to its probability distribution rule is processed, and avoids setting up complicated stochastic model.Can simulate the randomness feature that wind-power electricity generation is exerted oneself and photovoltaic generation is exerted oneself to be changed, the difficulty that can reduce modeling and solve again.Wind speed and intensity of illumination are divided into N interval by the method, according to the probability density function of wind speed and intensity of illumination, calculates the probability in each interval.But the method is just equidistantly sampled to continuous nondeterministic statement amount (wind speed, intensity of illumination), and limited status number cannot reflect actual conditions more accurately.And the probability density function eigenwert of the method is generally done approximate treatment, acquired results and actual value deviation are larger.

Summary of the invention

The object of the present invention is to provide a kind of generated output evaluation method and system, to improve the accuracy of generated output estimation.

The invention provides a kind of generated output evaluation method based on High Order Moment, the method comprises:

Step S1, generates multiple sampling sample groups, and the sampling sample group of the feature deviate f minimum of definite and original time series data; Wherein,

$f={\mathrm{\ϵ}}_1|{\mathrm{\μ}}_x-{\mathrm{\μ}}_x^{\′}|+{\mathrm{\ϵ}}_2|{\mathrm{\σ}}_x^2-{\mathrm{\σ}}_x^{\′2}|+{\mathrm{\ϵ}}_3|K_x^3-K_x^{\′3}|+{\mathrm{\ϵ}}_4|K_x^4-K_x^{\′4}|+{\mathrm{\ϵ}}_5|H_x^s-H_x^{\′s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of original time series data; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of a sampling sample group; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Step S2, according to the corresponding wind speed Weibull of sampling sample group probability distribution function simulation air speed data definite in step S1;

Step S3, estimates generated output according to the air speed data of simulation.

Preferably, described step S1 specifically comprises: generate multiple sampling sample groups, and adopt genetic algorithm to determine and the sampling sample group of the feature deviate f minimum of original time series data.

Preferably, described step S1 specifically comprises:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as X individuality of initial population, turns to step S12 afterwards;

Step S12, calculates Y the individuality that makes f minimum, turns to afterwards step S13;

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample sample group.

Preferably, described in, judge whether to reach to be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

Preferably, ε ₁, ε ₂, ε ₃, ε ₄, ε ₅value be followed successively by 0.35,0.25,0.15,0.15,0.1.

Preferably, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.

The present invention also provides a kind of generated output estimating system based on High Order Moment, comprising:

Sample group is chosen module, for generating multiple sampling sample groups, and the sampling sample group of the feature deviate f minimum of definite and original time series data; Wherein,

$f={\mathrm{\ϵ}}_1|{\mathrm{\μ}}_x-{\mathrm{\μ}}_x^{\′}|+{\mathrm{\ϵ}}_2|{\mathrm{\σ}}_x^2-{\mathrm{\σ}}_x^{\′2}|+{\mathrm{\ϵ}}_3|K_x^3-K_x^{\′3}|+{\mathrm{\ϵ}}_4|K_x^4-K_x^{\′4}|+{\mathrm{\ϵ}}_5|H_x^s-H_x^{\′s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of time series database; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of one group of sample; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Wind speed simulation module, for the sampling sample group corresponding wind speed Weibull probability distribution function simulation air speed data definite according to step S1;

Estimation block, estimates generated output for the air speed data of simulating according to described wind speed simulation module.

Preferably, sample group is chosen module specifically for carrying out:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as X individuality of initial population, turns to step S12 afterwards;

Step S12, calculates Y the individuality that makes f minimum, turns to afterwards step S13;

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample sample group.

Preferably, described in, judge whether to reach to be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

Preferably, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.

In generated output evaluation method provided by the invention, these low order squares of comprehensive mean and variance and these High Order Moment of the degree of bias, kurtosis and information entropy are simulated wind speed/intensity of illumination data, it is more accurate to make the simulation of wind speed/intensity of illumination data, thereby improves the levels of precision of the generated output estimation to wind-power electricity generation or photovoltaic generation.

Brief description of the drawings

The schematic flow sheet of a kind of generated output evaluation method based on High Order Moment that Fig. 1 provides for the embodiment of the present invention;

In a kind of generated output evaluation method based on High Order Moment that Fig. 2 provides for the embodiment of the present invention, adopt genetic algorithm to choose the schematic flow sheet of optimum sample group.

Embodiment

Below in conjunction with drawings and Examples, the specific embodiment of the present invention is further described.Following examples are only for technical scheme of the present invention is more clearly described, and can not limit the scope of the invention with this.

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.Following examples are used for illustrating the present invention, but do not limit the scope of the invention.

The embodiment of the present invention provides a kind of generated output evaluation method based on High Order Moment, and as shown in Figure 1, the method comprises:

Step S1, generates multiple sampling sample groups, and the sampling sample group of the feature deviate f minimum of definite and original time series data; Wherein,

$f={\mathrm{\ϵ}}_1|{\mathrm{\μ}}_x-{\mathrm{\μ}}_x^{\′}|+{\mathrm{\ϵ}}_2|{\mathrm{\σ}}_x^2-{\mathrm{\σ}}_x^{\′2}|+{\mathrm{\ϵ}}_3|K_x^3-K_x^{\′3}|+{\mathrm{\ϵ}}_4|K_x^4-K_x^{\′4}|+{\mathrm{\ϵ}}_5|H_x^s-H_x^{\′s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of original time series data; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of a sampling sample group; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Step S2, according to the corresponding wind speed Weibull of sampling sample group probability distribution function simulation air speed data definite in step S1.

General, the air speed data in a period of time generally all meets wind speed Weibull probability distribution function, and original time series data meets wind speed Weibull probability distribution function substantially, with it

The process of step S2 can be consistent according to the method for wind speed Weibull probability distribution function simulation air speed data with prior art, and the emphasis that the present invention pays close attention to does not lie in this, is not also elaborated at this.

Step S3, estimates generated output according to the air speed data of simulation.

Preferably, described step S1 specifically comprises: generate multiple sampling sample groups, and adopt genetic algorithm to determine and the sampling sample group of the feature deviate f minimum of original time series data.

Preferably, described step S1 specifically comprises:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as initial population, turns to step S12 afterwards.The X is here preset value, and concrete value can be set arbitrarily as required.

In practical application, the shape of weibull probability distribution function depends on its form parameter and scale parameter, can generate difform weibull probability distribution function by the corresponding form parameter of random generation and scale parameter.Afterwards, difform weibull probability distribution function is sampled, generate multiple different sampling sample groups.Each sampling sample group, as body one by one, joins in colony.

Step S12, calculates Y the colony that makes f minimum, turns to afterwards step S13.Here Y is more than or equal to 1 preset value.

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample group.

Be understood that, the selection here, heredity, variation are some individualities that eliminate in colony, and regenerate during sample group adds in colony as new individuality.

Preferably, described in, judge whether to reach to be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

The default evolutionary generation here refers to a default fixed value, and concrete value can be set arbitrarily as required.

Preferably, ε ₁, ε ₂, ε ₃, ε ₄, ε ₅value be followed successively by 0.35,0.25,0.15,0.15,0.1.

Preferably, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.

In the present invention, related X, Y, Z are the integer that is greater than 1, and X > Y, and occurrence can be set arbitrarily as required.

Below generated output evaluation method provided by the invention is further detailed:

Generated output evaluation method provided by the invention can be used for wind-power electricity generation power to estimate, also can estimate the power of photovoltaic generation, it is in the time setting up wind speed/illumination probability distribution function, introduce the high-order statistic such as average, variance, mean square value, the degree of bias, kurtosis, information entropy of wind speed/intensity of illumination, thereby make the more realistic wind speed/illumination probability distribution of wind speed/illumination probability distribution function of setting up, thereby make the generated output function of foundation more accurate.

Concrete, method provided by the invention can comprise following four parts;

One, the probability statistics of wind speed/intensity of illumination sampling

The probability statistics sampling of wind speed

Wind speed time series (v ₁, v ₂..., v _n) (k, Weibull c) distribute, and its probability density function is generally to obey two parameters

$f_v\left(v\right)=\frac{k}{c}{\left(\frac{v}{c}\right)}^{k-1}{e}^{[-{\left(\frac{v}{c}\right)}^k]}---\left(1\right)$

In formula: v is wind speed; C is scale parameter, has embodied the mean wind speed of this area's wind energy turbine set; K is form parameter, has reflected the characteristic of wind speed profile, and span is conventionally between 1.8 to 2.3.There are different scale parameters and shape coefficient in different regions.

Try to achieve its cumulative distribution function F by Weibull probability density function _v(v):

$F_v\left(v\right)={\∫}_0^v{f}_v\left(x\right)\mathrm{dx}---\left(2\right)$

Produce 0 to 1 random number sequence { x ₁, x ₂..., x _n, wind series corresponding to this random number sequence is

$v_i=F_v^{-1}\left(x_i\right)---\left(3\right)$

In formula: v _ibe i the wind speed that random number is corresponding; for F _v(v) inverse function; I=1,2 ..., n.Can obtain the time series data of 8760 periods of one year by the probability statistics sampling to wind speed.

The probability statistics sampling of intensity of illumination

The output power of photovoltaic array is main relevant with intensity of illumination and temperature.Consider that temperature impact is less, can think that exerting oneself of photovoltaic depends primarily on intensity of illumination.According to statistics, in certain hour section, Intensity of the sunlight can be similar to and regard Beta distribution as, and its probability density function is as follows:

$f_r\left(r\right)=\frac{\mathrm{\Γ}(\mathrm{\α}+\mathrm{\β})}{\mathrm{\Γ}\left(\mathrm{\α}\right)\mathrm{\Γ}\left(\mathrm{\β}\right)}\·{\left(\frac{r}{r_{\max }}\right)}^{\mathrm{\α}-1}\·{(1-\frac{r}{r_{\max }})}^{\mathrm{\β}-1}---\left(4\right)$

In formula: r and r _maxbe respectively actual light intensity and largest light intensity in the corresponding time period; α and β are the form parameter that Beta distributes; Γ is Gamma function.

Try to achieve its cumulative distribution function F by Beta probability density function _r(r):

$F_r\left(r\right)={\∫}_0^r{f}_r\left(x\right)\mathrm{dx}---\left(5\right)$

Produce 0 to 1 random number sequence { y ₁, y ₂..., y _n, intensity of illumination sequence corresponding to this random number sequence is

$r_i=F_r^{-1}\left(y_i\right)---\left(6\right)$

In formula: y _ibe i the intensity of illumination that random number is corresponding; for F _r(r) inverse function.Can obtain the time series data of 8760 periods of one year by the probability statistics sampling to light intensity.

Two, set up the computation model of feature deviate

Be principle according to each statistical nature and wind speed (intensity of illumination) the original time series data statistical characteristics deviation minimum of wind speed (intensity of illumination) probability model sampling sample, the computation model of setting up wind speed profile parameter k and scale coefficient c, intensity of illumination form parameter α and β is as follows:

$\begin{array}{c}f={\mathrm{\ϵ}}_1\mathrm{\Δ}{\mathrm{\μ}}_x+{\mathrm{\ϵ}}_2\mathrm{\Δ}{\mathrm{\σ}}_x^2+{\mathrm{\ϵ}}_3\mathrm{\Δ}{K}_x^3+{\mathrm{\ϵ}}_4\mathrm{\Δ}{K}_x^4+{\mathrm{\ϵ}}_5\mathrm{\Δ}{H}_x^s\\ ={\mathrm{\ϵ}}_1|{\mathrm{\μ}}_x-{\mathrm{\μ}}_x^{\′}|+{\mathrm{\ϵ}}_2|{\mathrm{\σ}}_x^2-{\mathrm{\σ}}_x^{\′2}|+{\mathrm{\ϵ}}_3|K_x^3-K_x^{\′3}|+{\mathrm{\ϵ}}_4|K_x^4-K_x^{\′4}|+{\mathrm{\ϵ}}_5|H_x^s-H_x^{\′s}|\end{array}---\left(7\right)$

In formula: μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of wind speed (intensity of illumination) original time series; μ ' _x, with be respectively the average, variance, the degree of bias, kurtosis and the information entropy that form sampling sample according to Weibull distribution (Beta distribution) sampling; Δ μ _x, with be respectively the deviation of wind speed (intensity of illumination) original time series and probability sampling sample average, variance, the degree of bias, kurtosis and information entropy; ε ₁, ε ₂, ε ₃, ε ₄and ε ₅be respectively the weight of average, variance, the degree of bias, kurtosis and information entropy, value is followed successively by 0.35,0.25,0.15,0.15,0.1.

Wherein, the account form of above-mentioned parameters can comprise:

(1) average

If x (n)={ x ₁, x ₂..., x _nbe one group of random signal, and the probability density of xi is pi, the first moment about the origin of stochastic variable x is defined as mathematical expectation:

$\mathrm{\μ}=E\left[x\left(n\right)\right]={\∫}_{-\∞}^{+\∞}{x}_i{p}_i\mathrm{dx}---\left(8\right)$

For time-limited stationary random signal sequence x (n), the reckoner of assembly average is shown:

${\mathrm{\μ}}_x=E\left[x\left(n\right)\right]=\frac{1}{N}\underset{n=1}{\overset{N}{\mathrm{\Σ}}}x\left(n\right)---\left(9\right)$

In formula: the sign of operation that E is expectation value; μ _xfor the oscillation centre that all value representation x (n) was worth in each moment.

(2) variance

In order to represent that the sampling value of stationary signal departs from its average value mu _xdegree, measure it in the upper and lower fluctuations of average, variance can be expressed as:

${{\mathrm{\σ}}_x}^2=E\left[{|x\left(n\right)-{\mathrm{\μ}}_x|}^2\right]---\left(10\right)$

For time-limited stationary signal sequence x (n), the calculating unbiased variance that its variance is estimated and have partial variance to be respectively:

${\mathrm{\σ}}_x^2=\frac{1}{N-1}\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{(x\left(n\right)-{\mathrm{\μ}}_x)}^2---\left(11\right)$

${\mathrm{\σ}}_x^2=\frac{1}{N}\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{(x\left(n\right)-{\mathrm{\μ}}_x)}^2---\left(12\right)$

(3) mean square value

The mean square value of continuous and stable signal, second order centre distance is defined as:

$D_x^2=E\left[{(x_i-0)}^2\right]={\∫}_{-\∞}^{+\∞}{x}_i^2{p}_i\mathrm{dx}---\left(13\right)$

Mean square value has reflected that signal x (n) changes in the average power in each moment, closes to be between the mean square value of discrete signal and variance:

$D_x^2={\mathrm{\σ}}_x^2+{\mathrm{\μ}}_x^2---\left(14\right)$

From above formula, the mean square value of signal can be tried to achieve by its variance and average.

(4) degree of bias

The degree of bias is the asymmetric degree of random signal variable with respect to its average, is a dimensionless numerical value based on three rank statistics, and it is defined as

$K_x^3={\∫}_{-\∞}^{+\∞}{(x-{\mathrm{\μ}}_x)}^3{p}_i\mathrm{dx}---\left(15\right)$

In actual applications, in the time that signal distributions approaches gaussian signal, its distribution is symmetrical, and the computing formula that has the degree of bias of limit for length's stationary sequence x (n) to estimate is:

$K_x^3=\sqrt{\frac{1}{6N}}\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{\left(\frac{x\left(n\right)-{\mathrm{\μ}}_x}{{\mathrm{\σ}}_x}\right)}^3---\left(16\right)$

The degree of bias is greater than zero for long streaking positively biased on the right, is less than zero negative bias for long streaking on the left side; Its absolute value is larger, and distributions shift degree is larger, if absolute value close to zero, distributional pattern is more close to just too distributing.

(5) kurtosis

Kurtosis is the characteristic quantity forming by the quadravalence centre distance of signal, and for the dimensionless group of the small impact comparison of ingredients sensitivity existing in signal, it is defined as:

$K_x^4=\sqrt{\frac{1}{24N}}[\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{\left(\frac{x\left(n\right)-{\mathrm{\μ}}_x}{{\mathrm{\σ}}_x}\right)}^4-3]---\left(17\right)$

If kurtosis value is regular representation, distribution is relatively sharp-pointed, and kurtosis value is the negative distribution relatively flat that represents.

(6) entropy

The concept of entropy was proposed by roentgen Rudolf Clausius first in 1850 at first, was used for representing the degree of uniformity that a kind of energy distributes in space, and energy distribution is more even, and entropy is just larger.In the time that the energy in this system is uniformly distributed completely, the entropy of system just reaches maximal value.The father Shannon of information theory has proposed the concept of information entropy in the paper of delivering for 1948, statistical entropy is applied in information theory as basic ingredient, call " information entropy " having got rid of the average information after redundancy in information, and provided the mathematic(al) representation of computing information entropy.In information theory, utilize quantity of information to represent that probabilistic size appears in message, it is defined as I=-log p, but quantity of information can not be used as whole system information measure, therefore on quantity of information basis, defines information entropy by mathematical expectation.

Be provided with sequence of discrete random variables x (n)={ x ₁, x ₂..., x _n, institute's given information source probability of occurrence is { p ₁, p ₂..., p _nand its information entropy can be expressed as:

$H=-\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{p}_i\log p_i---\left(18\right)$

In formula: the unit that has determined quantity of information at the bottom of logarithm, taking 2 as the unit of truth of a matter quantity of information be bit, Nat taking e as the unit of truth of a matter quantity of information, taking 10 as the unit of truth of a matter quantity of information be Hart, it is the numerical value that characterization information source overall information is estimated average meaning.

It is less that the different spectrum entropy of time domain is applicable to sampling number, contains the signal analysis method of noise sequence in synchronous signal.By discrete signal x (n)={ x who gathers ₁, x ₂..., x _n, taking length as M, the window that delay constant is 1, is divided into the mode data of N-M section by its order, and the mode data matrix having formed is decomposed.Can obtain singular value spectrum δ (i=1,2 ..., M) in nonzero eigenvalue more, the frequency distribution of signal is more complicated, if δ ₁>=δ ₂>=δ _m, known δ _ithat the one in time domain is divided to signal, the definition according to time-domain information entropy:

$\left\{\begin{array}{c}{H}_x^s=-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{p}_i\log p_i\\ p_i={\mathrm{\δ}}_i/\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\δ}}_i\end{array}\right.---\left(19\right)$

In formula: pi is that i singular value organized distribution probability in whole singular value spectrum ratio or i pattern in whole pattern.

Three, adopt genetic algorithm to solve and determine optimum sample group computation model

The sampling sample that the present invention obtains taking above-mentioned Part I is reference value, feature deviation is adaptive value, the form parameter k (wind speed) that Weibull distributes, form parameter α, β that scale parameter c and Beta distribute are that optimized variable (intensity of illumination) is set up mathematical model, adopt genetic algorithm to solve.

(1) chromosome coding and genetic manipulation

The present invention adopts the operation of binary segmentation formula chromosome.The operation of sectional type chromosome refers to carries out sectional type coding, sectional type intersection and sectional type variation to chromosome, as shown in Figure 1.

Sectional type coding is that whole chromosome is logically divided into two subsegments, and wherein first subsegment realizes the gene code to wind speed Weibull distribution shape parameter k and scale parameter c, is expressed as the coding of T1 and T2.Second subsegment realizes the gene code to Beta distribution shape parameter alpha, β, is expressed as the coding of S1 and S2.As shown in Fig. 2 (a).

Sectional type is intersected for generation of new chromosome.In the time intersecting, it is that three chromosomes of two parent chromosome x i, Xj are carried out respectively to an intersection.In Fig. 2 (b), the gene T21 of Weibull distribution parameter section intersects with T22, the gene S21 of Beta distribution parameter section intersects with S22.The new chromosome intersecting to form must meet the constraint condition of point fragment gene.

It is similar that sectional type variation and sectional type are intersected, and in the time carrying out mutation operation, two subsegments are carried out basic bit mutation operation (as Fig. 2 (c)) successively, and the result that makes a variation also must meet the constraint condition of point fragment gene.

Initial population produces from candidate solution space by random fashion, and the present invention gets 100; Require initial population to there is unicity, in initial population, there is no identical individuality; Each individuality should meet all constraint condition, to ensure that the individuality of selecting is efficient solution.

Fitness is to weigh individual good and bad unique index, is the key factor that concerns genetic algorithm success or failure.Can choosing of fitness function directly have influence on the speed of convergence of genetic algorithm and find optimum solution.It is deviation minimum that the present invention gets fitness, obtains the good and bad degree of each individuality.

Select operation also referred to as replicate run, be according to individual good and bad degree determine it the next generation by heredity or be eliminated.It is relatively larger that adaptable species are genetic to follow-on chance, and the weak species of adaptive faculty will be eliminated gradually.The present invention adopts elitism strategy method, and the method can be accelerated algorithm the convergence speed and improve the quality of separating.

The present invention arranges two end conditions: 1) reached the maximum evolutionary generation MaxGen of appointment, 2) the continuous Gen of optimal solution set generation all do not change.It is 50 that the present invention gets MaxGen; Gen is 5.

(2) algorithm execution step

1) input network data and initial calculation parameter, determines the form parameter k that Weibull distributes, form parameter α, β solution space that scale parameter c and Beta distribute.

2) sectional type coding generates initial population, and the initial population generating must meet constraint condition.

3) determine individual quality: the ideal adaptation degree value of calculating each objective function.

4) genetic manipulation forms a new generation's individuality: with the selection of elite's retention strategy, and make a variation and produce newly individuality by sectional type intersection and sectional type.

5) carry out end condition judgement: two end conditions are set: 1. reached the maximum evolutionary generation MaxGen of appointment, or 2. all do not change continuous Gen generation of optimal solution set.If meet any condition, output optimal case.Otherwise go to step 3).

Four wind-force, photovoltaic generation power probability model are asked for

Obtained after wind speed profile by above-mentioned Part III, just can obtain output power stochastic distribution by the approximation relation between output power of wind power generation and wind speed.Wind-power electricity generation active power of output P _wand the functional relation between wind speed v is:

$P_w=\left\{\begin{array}{cc}0& 0\&le;v<v_{\mathrm{ci}}\\ k_{1}v+k_2& v_{\mathrm{ci}}\&le;v<v_r\\ P_r& v_r\&le;v<v_{\mathrm{co}}\\ 0& v_{\mathrm{co}}\&le;v\end{array}\right.---\left(20\right)$

In formula: P _rfor aerogenerator rated power; v _cifor incision wind speed; v _rfor wind rating; v _cofor cut-out wind speed.And k ₁=P _rv _ci/ (v _ci-v _r) and k ₂=P _r/ (v _r-v _ci) be constant coefficient.

Obtain, after intensity of illumination distribution, just can obtaining by approximation relation between the output power of solar power system and light intensity the stochastic distribution of output power by 4.4.Suppose a given solar cell array, have M battery component, the area of each assembly and photoelectric transformation efficiency are for Wei not A _m, η _m, m=l, 2 ..., M, so the total output power of this solar cell array is:

P _m＝rAη (21)

In formula: A is the square formation total area; η is the total photoelectric transformation efficiency of square formation.They can be obtained by following formula:

$A=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{A}_m---\left(22\right)$

$\mathrm{\&eta;}=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{A}_m{\mathrm{\&eta;}}_m}{A}---\left(23\right)$

The probability density function of known light intensity, the probability density function that can obtain solar cell array output power also becomes Beta to distribute:

$f\left(P_m\right)=\frac{\mathrm{\&Gamma;}(\mathrm{\&alpha;}+\mathrm{\&beta;})}{\mathrm{\&Gamma;}\left(\mathrm{\&alpha;}\right)\mathrm{\&Gamma;}\left(\mathrm{\&beta;}\right)}\&CenterDot;{\left(\frac{P_m}{R_m}\right)}^{\mathrm{\&alpha;}-1}\&CenterDot;{(1-\frac{P_m}{R_m})}^{\mathrm{\&beta;}-1}---\left(24\right)$

In formula: R _mfor square formation peak power output, R _m=A η η _max

Based on identical design, the present invention also provides a kind of generated output estimating system based on High Order Moment, and for realizing above-mentioned generated output evaluation method, the method comprises:

Sample group is chosen module, for for generating multiple sampling sample groups, and determines and the sampling sample group of the feature deviate f minimum of original time series data; Wherein,

$f={\mathrm{\&epsiv;}}_1|{\mathrm{\&mu;}}_x-{\mathrm{\&mu;}}_x^{\&prime;}|+{\mathrm{\&epsiv;}}_2|{\mathrm{\&sigma;}}_x^2-{\mathrm{\&sigma;}}_x^{\&prime;2}|+{\mathrm{\&epsiv;}}_3|K_x^3-K_x^{\&prime;3}|+{\mathrm{\&epsiv;}}_4|K_x^4-K_x^{\&prime;4}|+{\mathrm{\&epsiv;}}_5|H_x^s-H_x^{\&prime;s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of time series database; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of one group of sample; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Wind speed simulation module, for the sampling sample group corresponding wind speed Weibull probability distribution function simulation air speed data definite according to step S1;

Estimation block, estimates generated output for the air speed data of simulating according to described wind speed simulation module.

Preferably, sample group is chosen module, specifically for generating multiple sampling sample groups, and adopts genetic algorithm to determine and the sampling sample group of the feature deviate f minimum of original time series data.

Sample group is chosen module specifically for carrying out:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as X individuality of initial population, turns to step S12 afterwards;

Step S12, calculates Y the individuality that makes f minimum, turns to afterwards step S13;

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample sample group.

Preferably, described in, judge whether to reach to be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

Preferably, ε ₁, ε ₂, ε ₃, ε ₄, ε ₅value be followed successively by 0.35,0.25,0.15,0.15,0.1.

Preferably, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.

The above is only the preferred embodiment of the present invention; it should be pointed out that for those skilled in the art, do not departing under the prerequisite of the technology of the present invention principle; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (10)

1. the generated output evaluation method based on High Order Moment, is characterized in that, described method comprises:

Step S1, generates multiple sampling sample groups, and the sampling sample group of the feature deviate f minimum of definite and original time series data; Wherein,

$f={\mathrm{\&epsiv;}}_1|{\mathrm{\&mu;}}_x-{\mathrm{\&mu;}}_x^{\&prime;}|+{\mathrm{\&epsiv;}}_2|{\mathrm{\&sigma;}}_x^2-{\mathrm{\&sigma;}}_x^{\&prime;2}|+{\mathrm{\&epsiv;}}_3|K_x^3-K_x^{\&prime;3}|+{\mathrm{\&epsiv;}}_4|K_x^4-K_x^{\&prime;4}|+{\mathrm{\&epsiv;}}_5|H_x^s-H_x^{\&prime;s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of original time series data; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of a sampling sample group; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Step S2, according to the corresponding wind speed Weibull of sampling sample group probability distribution function simulation air speed data definite in step S1;

Step S3, estimates generated output according to the air speed data of simulation.

2. the method for claim 1, is characterized in that, described step S1 specifically comprises: generate multiple sampling sample groups, and adopt genetic algorithm to determine and the sampling sample group of the feature deviate f minimum of original time series data.

3. method as claimed in claim 2, is characterized in that, described step S1 specifically comprises:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as X individuality of initial population, turns to step S12 afterwards;

Step S12, calculates Y the individuality that makes f minimum, turns to afterwards step S13;

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample sample group.

4. method claimed in claim 3, is characterized in that, described in judge whether to reach and be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

5. the method for claim 1, is characterized in that, ε ₁, ε ₂, ε ₃, ε ₄, ε ₅value be followed successively by 0.35,0.25,0.15,0.15,0.1.

6. the method as described in claim 1-5 any one, is characterized in that, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.

7. a generated output estimating system, is characterized in that, comprising:

Sample group is chosen module, for generating multiple sampling sample groups, and the sampling sample group of the feature deviate f minimum of definite and original time series data; Wherein,

$f={\mathrm{\&epsiv;}}_1|{\mathrm{\&mu;}}_x-{\mathrm{\&mu;}}_x^{\&prime;}|+{\mathrm{\&epsiv;}}_2|{\mathrm{\&sigma;}}_x^2-{\mathrm{\&sigma;}}_x^{\&prime;2}|+{\mathrm{\&epsiv;}}_3|K_x^3-K_x^{\&prime;3}|+{\mathrm{\&epsiv;}}_4|K_x^4-K_x^{\&prime;4}|+{\mathrm{\&epsiv;}}_5|H_x^s-H_x^{\&prime;s}|$

Described μ _x, with be respectively average, variance, the degree of bias, kurtosis and the information entropy of time series database; μ _x', with be respectively average, variance, the degree of bias, kurtosis and the information entropy of one group of sample; ε ₁, ε ₂, ε ₃, ε ₄, ε ₅be default weighted value;

Wind speed simulation module, for the sampling sample group corresponding wind speed Weibull probability distribution function simulation air speed data definite according to step S1;

Estimation block, estimates generated output for the air speed data of simulating according to described wind speed simulation module.

8. system as claimed in claim 7, is characterized in that, sample group is chosen module specifically for carrying out:

Step S11, random X the sampling sample group that meets different weibull probability distribution functions that generate, as X individuality of initial population, turns to step S12 afterwards;

Step S12, calculates Y the individuality that makes f minimum, turns to afterwards step S13;

Step S13, judges whether to reach evolution termination condition, if not, select, heredity, variation step, generate colony of new generation, turn to afterwards step S12; If so, will in the Y obtaining in a step S12 individuality, make the minimum body one by one of f value as object sample sample group.

9. system as claimed in claim 8, is characterized in that, described in judge whether to reach and be specially: judge whether one of to meet the following conditions:

Whether the number of times that carries out hereditary variation reaches default evolutionary generation, or the Y that the makes f minimum individuality obtaining for Z time continuously does not change.

10. system as claimed in claim 9, is characterized in that, described air speed data replaces with intensity of illumination data; Wind speed weibull probability distribution function replaces with intensity of illumination Beta function.