CN103559561A - Super-short-term prediction method of photovoltaic power station irradiance - Google Patents

Super-short-term prediction method of photovoltaic power station irradiance Download PDF

Info

Publication number
CN103559561A
CN103559561A CN201310571072.7A CN201310571072A CN103559561A CN 103559561 A CN103559561 A CN 103559561A CN 201310571072 A CN201310571072 A CN 201310571072A CN 103559561 A CN103559561 A CN 103559561A
Authority
CN
China
Prior art keywords
amp
irradiance
ij
step
data
Prior art date
Application number
CN201310571072.7A
Other languages
Chinese (zh)
Other versions
CN103559561B (en
Inventor
李卫
席林
佘慎思
曾旭
杨文斌
Original Assignee
上海电气集团股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 上海电气集团股份有限公司 filed Critical 上海电气集团股份有限公司
Priority to CN201310571072.7A priority Critical patent/CN103559561B/en
Publication of CN103559561A publication Critical patent/CN103559561A/en
Application granted granted Critical
Publication of CN103559561B publication Critical patent/CN103559561B/en

Links

Abstract

The invention discloses a super-short-term prediction method of photovoltaic power station irradiance. The method includes the steps that irradiance data are extracted from a history database, data of a night time quantum are removed, corresponding extraterrestrial theoretical irradiance is calculated, data abnormal detection is carried out based on the preceding operations, and the data are normalized in the difference value ratio method of an extraterrestrial irradiance theoretical value and practical irradiance; a training sample set is extracted according to input and output dimensionality of a model; a model of an irradiance time sequence is built through an ANFIS, a the rule quantity and an initial parameter of the ANFIS model are determined in a subtractive clustering method, and a fuzzy model parameter is optimized in a counter propagation algorithm and a least square method; a prediction sample is input, and a prediction value is obtained through calculation; the prediction value is added to form a new sample set, and multiple steps of prediction are achieved in a cycling mode; counter normalization processing is carried out on the prediction value. Super-short-term prediction of the irradiance can be achieved only by means of a history irradiance time sequence, prediction accuracy is good and the method is easy to carry out.

Description

A kind of ultra-short term Forecasting Methodology of photovoltaic plant irradiance

Technical field

The present invention relates to a kind of ultra-short term Forecasting Methodology of photovoltaic plant irradiance.

Background technology

It is the effective means of utilizing sun power that large-scale photovoltaic generates electricity by way of merging two or more grid systems, but sun power is a kind of intermittence, the undulatory property energy, and large-scale access meeting is to the safety of electric system, stable operation and guarantee that the quality of power supply brings severe challenge.If can make prediction more accurately to photovoltaic power station power generation power, can contribute to dispatching of power netwoks department formulate in time the rational method of operation and adjust exactly operation plan, can effectively alleviate the impact of photovoltaic electric station grid connection on whole electrical network, thereby guarantee the security and stability of electric system.

Affect photovoltaic power station power generation power meteorologic factor have multiple, as irradiation intensity, environment temperature and humidity etc.Wherein, irradiance has the greatest impact.And due to earth's surface irradiance, be subject to the impact of the multiple meteorologic factors such as cloud cover, temperature, air pressure, and there is very strong randomness, realize accurate prediction difficulty very large.For the ultra-short term prediction of ground irradiance, the method conventionally adopting has: (1) autoregressive moving average (ARMA) model.The method is utilized the historical time sequence data of irradiance, through Model Identification, parameter estimation, model testing, determine that can be described an irradiance seasonal effect in time series mathematical model, and then reach prediction object, but because ARMA is linear model, so precision of prediction is limited; (2) utilize the predicted method of sky cloud atlas image.The method conventionally need to be at the on-the-spot sky imager of installing of photovoltaic plant, and cost is expensive, and prediction algorithm is complicated, therefore implements comparatively difficulty.Therefore, how to realize the ultra-short term prediction of high-precision photovoltaic plant irradiance, become applicant and be devoted to the problem solving.

Summary of the invention

The object of the present invention is to provide a kind of ultra-short term Forecasting Methodology of photovoltaic plant irradiance, only need the ultra-short term prediction that utilizes historical irradiance time series to realize irradiance, precision of prediction is good, and easy to implement, for realizing the ultra-short term prediction of photovoltaic power station power generation power, lays the foundation.

The technical scheme that realizes above-mentioned purpose is:

A ultra-short term Forecasting Methodology for irradiance, comprises the following steps:

Step S1, from historical data base, extract the on-site actual irradiance data of section photovoltaic plant preset time, and weed out the invalid irradiation data of time period at night, according to the theoretically external irradiation degree of same time with area, the actual irradiance data of extracting is carried out to abnormal test, the irradiance data completing after abnormal test is normalized;

Step S2, determines training sample set by the input and output dimension of forecast model;

Step S3, adopts ANFIS to carry out modeling to irradiance time series, and model is 5 dimension input 1 dimension outputs, adopts subtractive clustering to determine regular number and the initial parameter of ANFIS model, and adopts back-propagation algorithm and least square method Optimization of Fuzzy model parameter;

Step S4, input prediction sample, calculates predicted value;

Step S5, judges that whether multi-step prediction completes, and if so, enters step S7; If not, enter step S6;

Step S6, adds predicted value to form new sample set, and returns to step S2;

Step S7, carries out renormalization processing by predicted value.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S1 comprises:

Step S11 extracts the on-site irradiance time sequential value of the photovoltaic plant v (t) of to be predicted day h days before as original sample collection from historical data base, and h is positive integer;

Step S12, calculates with the theoretically external irradiation degree of time with area;

Whether step S13, check each irradiance time sequential value v (t) normal according to formula: 0≤v (t)≤E (t), and E (t) is the t degree of external irradiation theoretically of this area constantly; If note abnormalities irradiance data, enter step S14; If not, enter step S15;

Step S14, rejects all irradiance data containing the data day of abnormal exposure degrees of data;

Step S15, according to formula:

x ( t ) = E ( t ) - v ( t ) E ( t ) ( E ( t ) ≠ 0 ) 0 ( E ( t ) = 0 )

Be normalized, x (t) is the data after normalization, 0≤x (t)≤1.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S12 utilizes following formula to calculate with the theoretically external irradiation degree of time with area:

Wherein, E is extraterrestrial irradiance value; E scit is solar constant; (r 0/ r) 2it is solar distance correction factor; δ is declination angle; E tit is the time difference; θ is a day angle; N is day of year; △ N is the value of correcting of day of year; Y refers to the time; INT refers to get the integral part of bracket inner digital; τ puts the solar hour angle in the required moment for this; S dit is the local time of this point; geographic latitude for this point; γ is the geographic longitude of this point; S is the hourage of this local meam time, and F is the number of minutes of this local meam time.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S2 refers to:

By the N+5 after normalization continuous irradiance time series { x 1, x 2, x 3..., x n+5resolve into the vector { V of N+1 5 dimensions 1..., V n+1, N is positive integer:

(x 1,x 2,x 3,x 4,x 5)=V 1

(x 2,x 3,x 4,x 5,x 6)=V 2

......

(x N,x N+1,x N+2,x N+3,x N+4)=V N

(x N+1,x N+2,x N+3,x N+4,x N+5)=V N+1

And then irradiance value and this vector in next moment of last dimension data of top n vector are matched, form N training sample pair: { (V 1, x 6), (V 2, x 7) ..., (V n, x n+5).

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S3 comprises:

Step S31, adopts ANFIS to carry out modeling to irradiance time series:

x ( t ) = ( Σ i = 1 n [ Σ j = 1 5 λ j i x ( t - j ) + ξ i ] · exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] ) ( Σ i = 1 n exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] )

Wherein, i=1,2 ..., n; J=1,2 ..., 5; X (t-j) is input quantity; X (t) is output quantity; ξ ifor consequent parameter, n is regular number; c ij, σ ijfor former piece parameter;

Step S32, carries out subtractive clustering analysis for training sample set, specifically refers to:

According to formula: calculate data point density D p,

Wherein, p, q=(1,2 ..., m), Y is sample pair, m is the right number of sample, δ afor the effective radius of neighbourhood of cluster centre;

Select density index mxm. obtain first cluster centre re-construct density function:

D ′ p = D p - D c 1 exp [ - | | Y p - y c 1 | | 2 ( δ b / 2 ) 2 ]

Wherein, δ b=1.25 δ a, utilize new density function to obtain the density index of all data points, determine next cluster centre again construct new density function, repeat this process until meet it is the density index mxm. of p cluster centre;

Thereby obtain optimum fuzzy rule number n and initial model former piece parameter c ijand σ ij;

Step S33, adopts blended learning method Optimized model parameter, and consequent adopts least squares identification parameter, and former piece adopts back-propagation algorithm Optimal Parameters:

Adopt least squares identification consequent parameter and ξ i: the fortran that step S31 is obtained is X=Φ θ, the matrix that Φ is m * 2n, the consequent parameter vector that θ is 2n * 1; X is the output vector of m * 1;

Make error criterion function be for desired output, according to principle of least square method, make J (θ) minimum, must have: thereby obtain optimized model consequent parameter and ξ i;

Fixing consequent parameter and ξ i, adopt back-propagation algorithm to adjust former piece parameter c ijand σ ij, correcting algorithm is:

c ij ( r + 1 ) = c ij ( r ) - α c ∂ E ∂ c ij σ ij ( r + 1 ) = σ ij ( r ) - α σ ∂ E ∂ σ ij

Wherein, α cand α σfor learning rate; c ij(r+1), σ ij(r+1), c ij(r), σ ij(r) represent respectively Center Parameter and the width parameter of the former piece membership function of the step of r+1 in correcting algorithm and r step.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein,

Described c ij ( r + 1 ) = c ij ( r ) - α c ∂ E ∂ c ij σ ij ( r + 1 ) = σ ij ( r ) - α σ ∂ E ∂ σ ij In,

Learning rate α cand α σinitial value get 0.01, number cycle of training of blended learning is 25,0≤r≤25.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S4 specifically refers to:

The optimum fuzzy rule number n that judgement obtains, if regular number is one, adopts the method for continuing prediction;

If regular number is more than one, by forecast sample V n+1be input to model, obtain predicted value x n+6if: 0≤x n+6≤ 1, represent that this predicted value is effective, otherwise, still adopt the method for continuing prediction.

The ultra-short term Forecasting Methodology of above-mentioned a kind of photovoltaic plant irradiance, wherein, described step S7 specifically refers to:

According to formula: P (t)=E (t)-x (t) * E (t), the predicted value obtaining is carried out to renormalization processing, wherein, x (t) is predicted value, and P (t) is the data after renormalization, and E (t) is the t degree of external irradiation theoretically value constantly.

The invention has the beneficial effects as follows: the present invention utilizes the historical irradiance data of photovoltaic plant collection in worksite, adopt the difference rule of three of extraterrestrial irradiance theoretical value and actual irradiance to carry out the original irradiance data of normalization, with adaptive neuro-fuzzy inference system, set up forecast model, with multistep recycle design, realize ultra-short term prediction, can effectively complete the ultra-short term prediction of irradiance, precision of prediction is good, and be easy to implement, thereby provide the foundation and ensure for Accurate Prediction photovoltaic plant ultra-short term generated output.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the ultra-short term Forecasting Methodology of photovoltaic plant irradiance of the present invention;

Fig. 2 is ANFIS model structure figure of the present invention;

Fig. 3 is the membership function of input variable 1;

Fig. 4 is the membership function of input variable 2;

Fig. 5 is the membership function of input variable 3;

Fig. 6 is the membership function of input variable 4;

Fig. 7 is the membership function of input variable 5;

Fig. 8 is predicted value and actual comparison curve;

Fig. 9 is predicated error curve.

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.

Refer to Fig. 1, the ultra-short term Forecasting Methodology of photovoltaic plant irradiance of the present invention, comprises the following steps:

Step S1, from historical data base, extract the on-site actual irradiance data of section photovoltaic plant preset time, according to the theoretically external irradiation degree of same time with area, the actual irradiance data of extracting is carried out to abnormal test, the irradiance data completing after abnormal test is normalized; In the present embodiment, utilize the weather station of photovoltaic plant, the irradiance data of collection site, frequency acquisition is 30 seconds, and then counts the irradiance mean value of 15 minutes, is kept in historical data base; Step S1 specifically comprises:

Step S11, from historical data base, extract to be predicted day before the on-site irradiance time sequential value of the photovoltaic plant v (t) (t irradiance actual observation data constantly) of h days as original sample collection, h is positive integer, and in the present embodiment, h is 30 days; Because irradiance data is null value at night, therefore get 48 15 minutes mean values in the middle of every day, namely from early 6 to the irradiance data 6 of evenings, form continuous time series;

Step S12, is ignoring under the minor impact prerequisites such as earth's axis precession and Ghandler motion, utilizes following formula to calculate with the theoretically external irradiation degree of time with area:

Wherein, E is extraterrestrial irradiance value; E scbe solar constant, value is 1367W/m2; (r 0/ r) 2it is solar distance correction factor; δ is declination angle; E tbe the time difference, unit divides (min); θ is a day angle; N is day of year (serial number of date in 1 year, January 1, day of year was 0, the day of year in Dec 31 non-leap year is 364, the leap year is 365); △ N is the value of correcting of day of year; Y refers to the time; INT refers to get the integral part of bracket inner digital; τ puts the solar hour angle in the required moment for this; S dit is the local time of this point; geographic latitude for this point; γ is the geographic longitude of this point; S is the hourage of this local meam time (being Beijing time), and F is the number of minutes of this local meam time.Input latitude, longitude and the time at regional place, can obtain the degree of external irradiation theoretically in this moment of this area.

Whether step S13, check each irradiance time sequential value v (t) normal according to formula: 0≤v (t)≤E (t), and E (t) is the t degree of external irradiation theoretically of this area constantly; If note abnormalities irradiance data, enter step S14; If not, enter step S15;

Step S14, rejects all irradiance data containing the data day of abnormal exposure degrees of data;

Step S15, according to formula:

x ( t ) = E ( t ) - v ( t ) E ( t ) ( E ( t ) ≠ 0 ) 0 ( E ( t ) = 0 )

Irradiance data after abnormality processing is normalized, and x (t) is the data after normalization, 0≤x (t)≤1.

Step S2, determines training sample set by the input and output dimension of forecast model, that is:

By the N+5 after normalization continuous irradiance time series { x 1, x 2, x 3..., x n+5resolve into the vector { V of N+1 5 dimensions 1..., V n+1, N is positive integer:

(x 1,x 2,x 3,x 4,x 5)=V 1

(x 2,x 3,x 4,x 5,x 6)=V 2

......

(x N,x N+1,x N+2,x N+3,x N+4)=V N

(x N+1,x N+2,x N+3,x N+4,x N+5)=V N+1

And then irradiance value and this vector in next moment of last dimension data of top n vector are matched, form N training sample pair: { (V 1, x 6), (V 2, x 7) ..., (V n, x n+5); V n+1be the prediction input of model, x n+5next be constantly worth x n+6be the value that we need to predict.

Step S3, adopts ANFIS to carry out modeling to irradiance time series, and model is 5 dimension input 1 dimension outputs, adopts subtractive clustering to determine regular number and the initial parameter of ANFIS model, and adopts back-propagation algorithm and least square method Optimization of Fuzzy model parameter; Step S3 comprises:

Step S31, adopts ANFIS to carry out modeling to irradiance time series, as follows:

Build irradiance seasonal effect in time series ANFIS model framework:

if x ( t - 1 ) is A 1 i , x ( t - 2 ) is A 2 i , x ( t - 3 ) is A 3 i , x ( t - 4 ) is A 4 i , x ( t - 5 ) is A 5 i , Then x i ( t ) = λ 1 i x ( t - 1 ) + λ 2 i x ( t - 2 ) + λ 3 i x ( t - 3 ) + λ 4 i x ( t - 4 ) + λ 5 i x ( t - 5 ) + ξ i

Wherein, i=1,2 ..., n; J=1,2 ..., 5; X (t-1), x (t-2), x (t-3), x (t-4), x (t-5) are input quantities, ξ ifor consequent parameter, n is regular number; it is the fuzzy set of input quantity x (t-j);

Fuzzy set adopts Gauss's membership function to represent: wherein, represent degree of membership; Former piece parameter c ijand σ ijthe center and the width that represent respectively membership function;

Utilize 5 layers of ANFIS network structure shown in Fig. 2 to obtain fuzzy inference rule:

Wherein, x (t) is output quantity;

By substitution,

x ( t ) = Σ i = 1 n x i ( t ) Π j = 1 5 μ A j i Σ i = 1 n Π j = 1 5 μ A j i = Σ i = 1 n x i ( t ) exp ( - ( Σ j = 1 5 ( x - ( t - j ) - c ij ) 2 2 σ ij 2 ) ) ( Σ i = 1 n exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] )

Again according to fuzzy inference rule, by x i(t) substitution above formula has following expression:

x ( t ) = ( Σ i = 1 n [ Σ j = 1 5 λ j i x ( t - j ) + ξ i ] · exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] ) ( Σ i = 1 n exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] ) ;

Step S32, subtractive clustering is using each data point as possible cluster centre, and calculate this as the possibility of cluster centre according to each data point data point density around, overcome the deficiency that other clustering procedure calculated amount increase with exponential manner along with the dimension of problem.For training sample set, carry out subtractive clustering analysis, specifically refer to:

According to formula: calculate data point density D p,

Wherein, p, q=(1,2 ..., m), Y is sample pair, m is the right number of sample, δ afor the effective radius of neighbourhood of cluster centre, be a positive number, in the present embodiment, be set as 0.5;

Select density index mxm. obtain first cluster centre re-construct density function:

D ′ p = D p - D c 1 exp [ - | | Y p - y c 1 | | 2 ( δ b / 2 ) 2 ]

Wherein, δ b=1.25 δ a, utilize new density function to obtain the density index of all data points, determine next cluster centre again construct new density function, repeat this process until meet it is the density index mxm. of p cluster centre;

Thereby obtain optimum fuzzy rule number n and initial model former piece parameter c ijand σ ij;

Step S33, adopts least squares identification consequent parameter and ξ i, that is: fortran step S31 being obtained is X=Φ θ, the matrix that Φ is m * 2n, the consequent parameter vector that θ is 2n * 1; X is the output vector of m * 1;

Make error criterion function be for desired output, according to principle of least square method, make J (θ) minimum, must have: thereby obtain optimized model consequent parameter and ξ i;

Fixing consequent parameter and ξ i, adopt back-propagation algorithm to adjust former piece parameter c ijand σ ij, consider error criterion function , x i(t) be t current output constantly, be desired output, correcting algorithm is:

c ij ( r + 1 ) = c ij ( r ) - α c ∂ E ∂ c ij σ ij ( r + 1 ) = σ ij ( r ) - α σ ∂ E ∂ σ ij

Wherein, learning rate α cand α σinitial value get 0.01, number cycle of training of blended learning is 25; c ij(r+1), σ ij(r+1), c ij(r), σ ij(r) represent respectively Center Parameter and the width parameter of the former piece membership function of the step of r+1 in correcting algorithm and r step, 0≤r≤25.

Step S4, input prediction sample, calculates predicted value; Specifically refer to:

The optimum fuzzy rule number n that obtains of judgement, if regular number is one, adopts the method for continuing prediction, usings the irradiance value of current time as next predicted value constantly;

If regular number is more than one, by forecast sample V n+1be input to model, obtain predicted value x n+6if: 0≤x n+6≤ 1, represent that this predicted value is effective, otherwise, still adopt the method for continuing prediction.So far, Single-step Prediction completes.

Step S5, judges that whether multi-step prediction completes, and if so, enters step S7; If not, enter step S6;

Step S6, adds predicted value to form new sample set, and returns to step S2; Repeat, circulate and obtain multi-step prediction value according to this;

Step S7, carries out renormalization processing by predicted value, refers to:

According to formula: P (t)=E (t)-x (t) * E (t), the predicted value obtaining is carried out to renormalization processing, model predication value is reduced to actual value; Wherein, x (t) is predicted value, and P (t) is the data after renormalization, and E (t) is the t degree of external irradiation theoretically value constantly.

Below with a concrete case explanation:

Select the irradiance data of District of Shanghai photovoltaic plant as experimental verification object.Gather on April 18th, 2013 to May 17 the 15 minutes mean value of irradiance of totally 30 days as initial modeling data sample, totally 2880, get every day early 6 to the data between 6 of evenings as valid data, totally 1440.According to step S12, calculate the theoretical irradiance of equal time period, according to step S13, detect and wherein have two days to have bad data, remain 1344 irradiance effective values of 28 days after rejecting these data of two days.The data of getting first 20 days are set up ANFIS model as original training sample collection, predict the irradiance value of latter 8 days, and prediction per hour once, is predicted 4 15 minutes mean values at every turn.

First front 960 data are made to normalized, then its division is consisted of to 955 training samples pair, every pair of sample is to being 5 dimension input vectors, and 1 ties up output vector.Adopt subtractive clustering to determine that number of fuzzy rules is 5, then adopt back propagation and least square method to obtain the front and back part parameter of model.The membership function of five input quantities of ANFIS model is as shown in Fig. 3 to Fig. 7.In figure, in1ct1~in1ct4 represents four membership function curve maps of the first dimension input variable; In2ct1~in2ct4 represents four membership function curve maps of the second dimension input variable; In3ct1~in3ct4 represents four membership function curve maps of third dimension input variable; In4ct1~in4ct4 represents four membership function curve maps of fourth dimension input variable; In5ct1~in5ct4 represents four membership function curve maps of the 5th dimension input variable.

The parameter of fuzzy rule former piece Gauss membership function is as shown in table 1, and consequent linear function parameter is as shown in table 2:

Table 1

Fuzzy rule Consequent linear dimensions (λ 1,λ 2,λ 3,λ 4,λ 5,ξ) R1 [-0.082?-0.063?0.123?-0.169?1.244?-0.060] R2 [-0.171?0.174?-0.291?0.797?0.588?-0.028] R3 [2.334?-0.783?-1.247?0.982?-0.504?0.165] R4 [0.029?-0.089?0.155?0.026?0.974?-0.036] R5 [-0.076?0.277?-0.104?-0.055?1.104?-0.089]

Table 2

5 15 minutes actual irradiance mean values (being between 17:45 to 18:00 on May 9) before May 10 6:00 are as forecast sample input model, can draw the mean value of next 15 minutes (being between 6:00 to 6:15 on May 10), this predicted value is added to forecast sample, repeat the predicted value that above process can obtain following 1 hour (being between 6:00 to 7:00 on May 10) for 4 times.Carry out prediction once at interval of one hour, circulate and can obtain the irradiance predicted value of following 8 days for 96 times, totally 384 points, it is 59.7 that statistics obtains root-mean-square error (RMSE), standard root square error (NRMSE) is 0.4259.Wherein, RMSE and NRMSE are calculated as follows respectively:

RMSE = [ 1 N ′ Σ i = 1 N ′ ( P ^ i - P i ) 2 ] 1 / 2

NRMSE = ( 1 N ′ Σ i = 1 N ′ ( P ^ i - P i ) 2 ) 1 / 2 1 N ′ Σ i = 1 N ′ P i

In formula, P ifor prediction irradiance, for actual measurement irradiance, the number that N' is future position.

As shown in Figure 8, predicated error curve as shown in Figure 9, visible for predicted value and actual comparison curve, and forecast model possesses good precision of prediction.

Above embodiment is used for illustrative purposes only, but not limitation of the present invention, person skilled in the relevant technique, without departing from the spirit and scope of the present invention, can also make various conversion or modification, therefore all technical schemes that are equal to also should belong to category of the present invention, should be limited by each claim.

Claims (8)

1. a ultra-short term Forecasting Methodology for photovoltaic plant irradiance, is characterized in that, comprises the following steps:
Step S1, from historical data base, extract the on-site actual irradiance data of section photovoltaic plant preset time, and weed out the invalid irradiation data of time period at night, according to the theoretically external irradiation degree of same time with area, the actual irradiance data of extracting is carried out to abnormal test, the irradiance data completing after abnormal test is normalized;
Step S2, determines training sample set by the input and output dimension of forecast model;
Step S3, adopts ANFIS to carry out modeling to irradiance time series, and model is 5 dimension input 1 dimension outputs, adopts subtractive clustering to determine regular number and the initial parameter of ANFIS model, and adopts back-propagation algorithm and least square method Optimization of Fuzzy model parameter;
Step S4, input prediction sample, calculates predicted value;
Step S5, judges that whether multi-step prediction completes, and if so, enters step S7; If not, enter step S6;
Step S6, adds predicted value to form new sample set, and returns to step S2;
Step S7, carries out renormalization processing by predicted value.
2. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 1, is characterized in that, described step S1 comprises:
Step S11 extracts the on-site irradiance time sequential value of the photovoltaic plant v (t) of to be predicted day h days before as original sample collection from historical data base, and h is positive integer;
Step S12, calculates with the theoretically external irradiation degree of time with area;
Whether step S13, check each irradiance time sequential value v (t) normal according to formula: 0≤v (t)≤E (t), and E (t) is the t degree of external irradiation theoretically of this area constantly; If note abnormalities irradiance data, enter step S14; If not, enter step S15;
Step S14, rejects all irradiance data containing the data day of abnormal exposure degrees of data;
Step S15, according to formula:
x ( t ) = E ( t ) - v ( t ) E ( t ) ( E ( t ) ≠ 0 ) 0 ( E ( t ) = 0 )
Be normalized, x (t) is the data after normalization, 0≤x (t)≤1.
3. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 2, is characterized in that, described step S12 utilizes following formula to calculate with the theoretically external irradiation degree of time with area:
Wherein, E is extraterrestrial irradiance value; E scit is solar constant; (r 0/ r) 2it is solar distance correction factor; δ is declination angle; E tit is the time difference; θ is a day angle; N is day of year; △ N is the value of correcting of day of year; Y refers to the time; INT refers to get the integral part of bracket inner digital; τ puts the solar hour angle in the required moment for this; S dit is the local time of this point; geographic latitude for this point; γ is the geographic longitude of this point; S is the hourage of this local meam time, and F is the number of minutes of this local meam time.
4. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 1, is characterized in that, described step S2 refers to:
By the N+5 after normalization continuous irradiance time series { x 1, x 2, x 3..., x n+5resolve into the vector { V of N+1 5 dimensions 1..., V n+1, N is positive integer:
(x 1,x 2,x 3,x 4,x 5)=V 1
(x 2,x 3,x 4,x 5,x 6)=V 2
......
(x N,x N+1,x N+2,x N+3,x N+4)=V N
(x N+1,x N+2,x N+3,x N+4,x N+5)=V N+1
And then irradiance value and this vector in next moment of last dimension data of top n vector are matched, form N training sample pair: { (V 1, x 6), (V 2, x 7) ..., (V n, x n+5).
5. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 1 and 2, is characterized in that, described step S3 comprises:
Step S31, adopts ANFIS to carry out modeling to irradiance time series:
x ( t ) = ( Σ i = 1 n [ Σ j = 1 5 λ j i x ( t - j ) + ξ i ] · exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] ) ( Σ i = 1 n exp [ - ( Σ j = 1 5 ( x ( t - j ) - c ij ) 2 2 σ ij 2 ) ] )
Wherein, i=1,2 ..., n; J=1,2 ..., 5; X (t-j) is input quantity; X (t) is output quantity; ξ ifor consequent parameter, n is regular number; c ij, σ ijfor former piece parameter;
Step S32, carries out subtractive clustering analysis for training sample set, specifically refers to:
According to formula: calculate data point density D p,
Wherein, p, q=(1,2 ..., m), Y is sample pair, m is the right number of sample, δ afor the effective radius of neighbourhood of cluster centre;
Select density index mxm. obtain first cluster centre re-construct density function:
D ′ p = D p - D c 1 exp [ - | | Y p - y c 1 | | 2 ( δ b / 2 ) 2 ]
Wherein, δ b=1.25 δ a, utilize new density function to obtain the density index of all data points, determine next cluster centre again construct new density function, repeat this process until meet it is the density index mxm. of p cluster centre;
Thereby obtain optimum fuzzy rule number n and initial model former piece parameter c ijand σ ij;
Step S33, adopts blended learning method Optimized model parameter, and consequent adopts least squares identification parameter, and former piece adopts back-propagation algorithm Optimal Parameters:
Adopt least squares identification consequent parameter and ξ i: the fortran that step S31 is obtained is X=Φ θ, the matrix that Φ is m * 2n, the consequent parameter vector that θ is 2n * 1; X is the output vector of m * 1;
Make error criterion function be for desired output, according to principle of least square method, make J (θ) minimum, must have: thereby obtain optimized model consequent parameter and ξ i;
Fixing consequent parameter and ξ i, adopt back-propagation algorithm to adjust former piece parameter c ijand σ ij, correcting algorithm is:
c ij ( r + 1 ) = c ij ( r ) - α c ∂ E ∂ c ij σ ij ( r + 1 ) = σ ij ( r ) - α σ ∂ E ∂ σ ij
Wherein, α cand α σfor learning rate; c ij(r+1), σ ij(r+1), c ij(r), σ ij(r) represent respectively Center Parameter and the width parameter of the former piece membership function of the step of r+1 in correcting algorithm and r step.
6. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 5, is characterized in that,
Described c ij ( r + 1 ) = c ij ( r ) - α c ∂ E ∂ c ij σ ij ( r + 1 ) = σ ij ( r ) - α σ ∂ E ∂ σ ij In,
Learning rate α cand α σinitial value get 0.01, number cycle of training of blended learning is 25,0≤r≤25.
7. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 1, is characterized in that, described step S4 specifically refers to:
The optimum fuzzy rule number n that judgement obtains, if regular number is one, adopts the method for continuing prediction;
If regular number is more than one, by forecast sample V n+1be input to model, obtain predicted value x n+6if: 0≤x n+6≤ 1, represent that this predicted value is effective, otherwise, still adopt the method for continuing prediction.
8. the ultra-short term Forecasting Methodology of a kind of photovoltaic plant irradiance according to claim 1, is characterized in that, described step S7 specifically refers to:
According to formula: P (t)=E (t)-x (t) * E (t), the predicted value obtaining is carried out to renormalization processing, wherein, x (t) is predicted value, and P (t) is the data after renormalization, and E (t) is the t degree of external irradiation theoretically value constantly.
CN201310571072.7A 2013-11-13 2013-11-13 A kind of ultra-short term prediction method of photovoltaic plant irradiance CN103559561B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310571072.7A CN103559561B (en) 2013-11-13 2013-11-13 A kind of ultra-short term prediction method of photovoltaic plant irradiance

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310571072.7A CN103559561B (en) 2013-11-13 2013-11-13 A kind of ultra-short term prediction method of photovoltaic plant irradiance

Publications (2)

Publication Number Publication Date
CN103559561A true CN103559561A (en) 2014-02-05
CN103559561B CN103559561B (en) 2016-03-09

Family

ID=50013802

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310571072.7A CN103559561B (en) 2013-11-13 2013-11-13 A kind of ultra-short term prediction method of photovoltaic plant irradiance

Country Status (1)

Country Link
CN (1) CN103559561B (en)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103823504A (en) * 2014-03-20 2014-05-28 常州工学院 Maximum power tracing control method based on least squares support vector machine
CN104933489A (en) * 2015-06-29 2015-09-23 东北电力大学 Wind power real-time high precision prediction method based on adaptive neuro-fuzzy inference system
CN105373857A (en) * 2015-11-27 2016-03-02 许昌许继软件技术有限公司 Photovoltaic power station irradiance prediction method
CN105590027A (en) * 2015-12-17 2016-05-18 国网冀北电力有限公司 Identification method for photovoltaic power abnormal data
CN105701572A (en) * 2016-01-13 2016-06-22 国网甘肃省电力公司电力科学研究院 Photovoltaic short-term output prediction method based on improved Gaussian process regression
CN105787594A (en) * 2016-02-29 2016-07-20 南京航空航天大学 Irradiation prediction method based on multivariate time series and regression analysis
CN105956685A (en) * 2016-04-20 2016-09-21 南京国电南自电网自动化有限公司 Photovoltaic power factor table prediction method
CN106295034A (en) * 2016-08-15 2017-01-04 河海大学常州校区 A kind of high accuracy scattering radiometer calculates method
CN106300424A (en) * 2016-09-07 2017-01-04 广东工业大学 A kind of method and device determining new forms of energy user's photovoltaic generating system daily generation
CN106651007A (en) * 2016-11-24 2017-05-10 北京理工大学 Method and device for GRU-based medium and long-term prediction of irradiance of photovoltaic power station
CN107171638A (en) * 2016-10-18 2017-09-15 上海电力新能源发展有限公司 The division methods and device of a kind of group of set of strings
CN107256437A (en) * 2017-05-15 2017-10-17 内蒙古电力(集团)有限责任公司 A kind of photovoltaic plant ultra-short term irradiation level Forecasting Methodology and system
CN107358323A (en) * 2017-07-13 2017-11-17 上海交通大学 A kind of power forecasting method of short-term photovoltaic generation
CN109447372A (en) * 2018-11-13 2019-03-08 广东电网有限责任公司 One kind is avoided the peak hour load forecasting method and device

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107085755A (en) * 2017-05-15 2017-08-22 内蒙古电力(集团)有限责任公司 A kind of photovoltaic plant short term power Forecasting Methodology and system

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090312999A1 (en) * 2008-06-17 2009-12-17 Kasztenny Bogdan Z Systems and methods for predicting maintenance of intelligent electronic devices
CN103019267A (en) * 2012-12-10 2013-04-03 华东交通大学 Predicative control method for modeling and running speed of adaptive network-based fuzzy inference system (ANFIS) of high-speed train

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090312999A1 (en) * 2008-06-17 2009-12-17 Kasztenny Bogdan Z Systems and methods for predicting maintenance of intelligent electronic devices
CN103019267A (en) * 2012-12-10 2013-04-03 华东交通大学 Predicative control method for modeling and running speed of adaptive network-based fuzzy inference system (ANFIS) of high-speed train

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A MELLIT,ET AL.: "An ANFIS-based Forecasting for Solar Radiation Data from Sunshine Duration and Ambient Temperature》", 《IEEE POWER ENGINEERING SOCIETY GENERAL MEETING》 *
周岩: "《基于多模型模糊神经网络的智能天气预报》", 《中国优秀硕士学位论文全文数据库基础科学辑(月刊)》 *

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103823504B (en) * 2014-03-20 2016-01-20 常州工学院 A kind of maximum power tracking and controlling method based on least square method supporting vector machine
CN103823504A (en) * 2014-03-20 2014-05-28 常州工学院 Maximum power tracing control method based on least squares support vector machine
CN104933489A (en) * 2015-06-29 2015-09-23 东北电力大学 Wind power real-time high precision prediction method based on adaptive neuro-fuzzy inference system
CN105373857A (en) * 2015-11-27 2016-03-02 许昌许继软件技术有限公司 Photovoltaic power station irradiance prediction method
CN105590027A (en) * 2015-12-17 2016-05-18 国网冀北电力有限公司 Identification method for photovoltaic power abnormal data
CN105701572A (en) * 2016-01-13 2016-06-22 国网甘肃省电力公司电力科学研究院 Photovoltaic short-term output prediction method based on improved Gaussian process regression
CN105787594A (en) * 2016-02-29 2016-07-20 南京航空航天大学 Irradiation prediction method based on multivariate time series and regression analysis
CN105787594B (en) * 2016-02-29 2019-11-12 南京航空航天大学 Irradiation prediction technique based on multivariate time series and regression analysis
CN105956685B (en) * 2016-04-20 2019-09-10 南京国电南自电网自动化有限公司 A kind of photovoltaic power factor table prediction technique
CN105956685A (en) * 2016-04-20 2016-09-21 南京国电南自电网自动化有限公司 Photovoltaic power factor table prediction method
CN106295034A (en) * 2016-08-15 2017-01-04 河海大学常州校区 A kind of high accuracy scattering radiometer calculates method
CN106295034B (en) * 2016-08-15 2019-12-03 河海大学常州校区 A kind of high-precision scattering radiometer calculation method
CN106300424A (en) * 2016-09-07 2017-01-04 广东工业大学 A kind of method and device determining new forms of energy user's photovoltaic generating system daily generation
CN106300424B (en) * 2016-09-07 2018-10-09 广东工业大学 A kind of method and device of determining new energy user photovoltaic generating system daily generation
CN107171638A (en) * 2016-10-18 2017-09-15 上海电力新能源发展有限公司 The division methods and device of a kind of group of set of strings
CN106651007A (en) * 2016-11-24 2017-05-10 北京理工大学 Method and device for GRU-based medium and long-term prediction of irradiance of photovoltaic power station
CN107256437A (en) * 2017-05-15 2017-10-17 内蒙古电力(集团)有限责任公司 A kind of photovoltaic plant ultra-short term irradiation level Forecasting Methodology and system
CN107358323A (en) * 2017-07-13 2017-11-17 上海交通大学 A kind of power forecasting method of short-term photovoltaic generation
CN109447372A (en) * 2018-11-13 2019-03-08 广东电网有限责任公司 One kind is avoided the peak hour load forecasting method and device

Also Published As

Publication number Publication date
CN103559561B (en) 2016-03-09

Similar Documents

Publication Publication Date Title
Wang et al. A novel hybrid system based on a new proposed algorithm—Multi-Objective Whale Optimization Algorithm for wind speed forecasting
Zhang et al. Modeling conditional forecast error for wind power in generation scheduling
Almonacid et al. A methodology based on dynamic artificial neural network for short-term forecasting of the power output of a PV generator
Zeng et al. Short-term solar power prediction using a support vector machine
Liu et al. A systematic approach for dynamic security assessment and the corresponding preventive control scheme based on decision trees
Sheng et al. Short-term solar power forecasting based on weighted Gaussian process regression
Chen et al. Solar radiation forecast based on fuzzy logic and neural networks
Shi et al. Forecasting power output of photovoltaic systems based on weather classification and support vector machines
Khatib et al. A review of solar energy modeling techniques
Ekici A least squares support vector machine model for prediction of the next day solar insolation for effective use of PV systems
Yona et al. Determination method of insolation prediction with fuzzy and applying neural network for long-term ahead PV power output correction
Stathopoulos et al. Wind power prediction based on numerical and statistical models
Qing et al. Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM
De Giorgi et al. Error analysis of short term wind power prediction models
Shi et al. Hybrid forecasting model for very-short term wind power forecasting based on grey relational analysis and wind speed distribution features
CN103353952B (en) A kind of photovoltaic power Forecasting Methodology based on ground cloud atlas
CN102663513B (en) Utilize the wind power combined prediction modeling method of grey relational grade analysis
Elminir et al. Prediction of hourly and daily diffuse fraction using neural network, as compared to linear regression models
Li et al. Estimating daily global solar radiation by day of year in China
Lam et al. Solar radiation modelling using ANNs for different climates in China
Rajanna et al. Modeling of integrated renewable energy system for electrification of a remote area in India
Bekele et al. Wind energy potential assessment at four typical locations in Ethiopia
Almonacid et al. Estimating the maximum power of a high concentrator photovoltaic (HCPV) module using an artificial neural network
Ghaderi et al. Deep forecast: Deep learning-based spatio-temporal forecasting
Koo et al. Estimation of the monthly average daily solar radiation using geographic information system and advanced case-based reasoning

Legal Events

Date Code Title Description
PB01 Publication
C06 Publication
SE01 Entry into force of request for substantive examination
GR01 Patent grant
C14 Grant of patent or utility model