CN102521670A - Power generation output power prediction method based on meteorological elements for photovoltaic power station - Google Patents

Abstract

The invention discloses a power generation output power prediction method based on meteorological elements for a photovoltaic power station. The prediction method comprises the following steps that: a, the history records of the meteorological element data of the location in which the photovoltaic power station is located and the output power corresponding to each record are obtained; b, the meteorological element data is corrected into immediate data of a photovoltaic panel; c, the corrected meteorological element data is taken as input data and is inputted to a BP (Back Propagation) neural network, and the output power corresponding to the meteorological element data is taken as the input of the BP neural network to train the BP neural network; d, the meteorological element data of the location of the photovoltaic power station in the prediction time slot is obtained according to the numerical weather prediction, and the meteorological element data is corrected into the immediate data of the photovoltaic panel to generate the corrected meteorological element data; and e, the corrected meteorological element data obtained in the step d is inputted to the BP neural network, and the output data is the power generation output power of the photovoltaic power station in the prediction time slot. The power generation output power prediction method based on meteorological elements for the photovoltaic power station is simple and feasible and is high in accuracy.

Description

Photovoltaic power station power generation output power Forecasting Methodology based on meteorological element

Technical field

The present invention relates to the photovoltaic power generation technology field, specifically a kind of photovoltaic power station power generation output power Forecasting Methodology based on meteorological element.

Background technology

Solar energy power generating is to utilize the photovoltaic effect of solar cell solar radiant energy directly to be converted into a kind of forms of electricity generation of electric energy.Present stage, applying of sun power presents world trends in the ascendant day by day, and solar energy industry becomes one of new forms of energy industry that is surging forward in the whole world.The sun power of development and use cleaning, safety, environmental protection becomes human society and alleviates the energy starved that increasingly sharpens and select and administer the effective strength of severe environmental pollution jointly.The stable operation of electrical network need keep certain balance between both sides of supply and demand, promptly change according to user's consumption, and preset the unlatching of genset such as thermoelectricity, water power and close down, thus the general power of adjustment supply correspondingly.Because photovoltaic generation receives the influence of weather bigger, and can not freely control as thermoelectricity and water power, so the output power of photovoltaic power station power generation has characteristics such as acute variation and intermittence.Thus, photovoltaic plant is connected to the grid and will the balance of electrical network be had an immense impact on.

1) peaking problem.Along with the variation of weather, the output power acute variation of photovoltaic plant has a strong impact on the peak regulation of electrical network;

2) stabilization of power grids problem.When big disturbance took place electrical network, photovoltaic plant was not owing to possess low voltage ride-through capability, thereby out of service easily electrical network brought secondary pulse, influenced the transient stability of electrical network;

So the photovoltaic plant output power is effectively monitored and is predicted, include the photovoltaic plant output power generation schedule establishment of electrical network in, and participate in Real-Time Scheduling, be one of important measures that guarantee stabilization of power grids economical operation.Automatically control thereby can implement generator operation, realize polynary power supply combined dispatching.

The research of solar energy power generating power prediction is started late.States such as Germany, Denmark, Japan, the U.S., France and Canada all carried out correlative study.Mainly be to set up the solar energy resources monitoring point in the whole country, collect the solar energy resources data, and set up the photovoltaic generation power prediction model, predict the spatial and temporal distributions that nationwide photovoltaic generation is exerted oneself.Though abroad carried out the solar energy resources correlative study of monitoring and photovoltaic plant power prediction system of layouting, but still belonged to the starting stage, also do not had the photovoltaic plant power prediction systems approach of maturation at present.

In view of this, the inventor actively studies and founds, and to invent a kind of photovoltaic power station power generation output power Forecasting Methodology based on meteorological element, realizes the accurate prediction to the photovoltaic plant output power.

Summary of the invention

In order to solve the problems referred to above that exist in the prior art, the invention provides a kind of photovoltaic power station power generation output power Forecasting Methodology based on meteorological element.The inventive method has simple, the characteristics that accuracy is high.

In order to solve the problems of the technologies described above, the present invention has adopted following technical scheme:

Photovoltaic power station power generation output power Forecasting Methodology based on meteorological element comprises the steps:

A. obtain the on-site historical record of the meteorological element data of irradiation intensity, temperature and wind speed that comprises of photovoltaic plant and reach the photovoltaic power station power generation output power relative with each record;

B. wherein, irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel, temperature is modified to the effective temperature of photovoltaic panel, thereby generates revised meteorological element data;

C. revised meteorological element data are imported the BP neural network as the input data, photovoltaic power station power generation output power that will be corresponding with each meteorological element data is trained the BP neural network as the input of BP neural network;

D. obtain the meteorological element data that comprise irradiation intensity, temperature and wind speed of photovoltaic plant location according to numerical weather forecast in the predicted time section; And irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel; Temperature is modified to the effective temperature of photovoltaic panel, thereby generates revised meteorological element data;

E. with the BP neural network of revised meteorological element data input after step c training of steps d gained, the data of BP neural network output are the generating output power of the photovoltaic plant of this predicted time section.

Further, in said step b and the steps d, irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel through following formula:

$I_t=I_b\cos {\mathrm{\θ}}_i+I_d\left(\frac{1+\cos \mathrm{\β}}{2}\right)+\mathrm{\ρ}{I}_h\left(\frac{1-\cos \mathrm{\β}}{2}\right)$

In the formula, I _tBe the effective irradiation intensity of photovoltaic panel, I _bBe direct projection irradiation intensity, I _dBe scattering irradiation intensity, I _hBe the total irradiation intensity of surface level, β is the photovoltaic panel inclination angle, θ _iBe solar incident angle, ρ is a reflection coefficient.

Further, in said step b and the steps d, temperature is modified to the effective temperature of photovoltaic panel: T=T through following formula _Air+ KS

In the formula, T is the effective temperature of photovoltaic panel, T _AirBe environment temperature, S is an intensity of illumination, and K is a temperature coefficient.

Further, the input data of BP neural network also comprise time data, said time data comprises month, day and the time.

Compared with prior art, beneficial effect of the present invention is:

Photovoltaic power station power generation output power Forecasting Methodology based on meteorological element of the present invention is according to the residing geographic position of photovoltaic plant; Having analyzed influences the various meteorologic factors that photovoltaic plant is exerted oneself; Utilize the historical numerical weather forecast and the output power of the photovoltaic plant of history to set up neural network model; Realization is to the prediction of following photovoltaic plant output power, and is simple.The inventive method has adopted the method for artificial intelligence, does not need each class feature of photovoltaic plant inner member, the error of having avoided the component parameters out of true to cause, and prediction effect is better, and accuracy is high.Be the cooperation of photovoltaic plant and conventional power supply, ensure providing the foundation property of the measures data of power network safety operation.

Embodiment

Below in conjunction with specific embodiment the present invention is described in further detail, but not as to qualification of the present invention.

Photovoltaic power station power generation output power Forecasting Methodology based on meteorological element comprises the steps:

A. obtain the on-site historical record of the meteorological element data of irradiation intensity, temperature and wind speed that comprises of photovoltaic plant and reach the photovoltaic power station power generation output power relative with each record; This moment or irradiation intensity be the irradiation intensity in the surface level, temperature is an environment temperature, so need step b that it is modified to effective irradiation intensity and effective temperature to photovoltaic panel respectively.

B. irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel, temperature is modified to the effective temperature of photovoltaic panel, thereby generate revised meteorological element data;

C. revised meteorological element data are imported the BP neural network as the input data, photovoltaic power station power generation output power that will be corresponding with each meteorological element data is trained the BP neural network as the input of BP neural network;

D. obtain the meteorological element data that comprise irradiation intensity, temperature and wind speed of photovoltaic plant location according to numerical weather forecast in the predicted time section; And irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel; Temperature is modified to the effective temperature of photovoltaic panel, thereby generates revised meteorological element data;

E. with the BP neural network of revised meteorological element data input after step c training of steps d gained, the data of BP neural network output are the generating output power of the photovoltaic plant of this predicted time section.

Because the irradiation intensity that step a and d obtain is the total irradiation intensity of surface level; And generated energy is directly related with the effective irradiation intensity of photovoltaic panel; So need irradiation intensity be modified to the effective irradiation intensity of photovoltaic panel, and the temperature that step a and d obtain is an environment temperature, and generated energy is directly related with the effective temperature of photovoltaic panel; So need temperature be modified to the effective temperature of photovoltaic panel, thereby generate revised meteorological element data.In step b and the steps d, irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel through following formula:

$I_t=I_b\cos {\mathrm{\θ}}_i+I_d\left(\frac{1+\cos \mathrm{\β}}{2}\right)+\mathrm{\ρ}{I}_h\left(\frac{1-\cos \mathrm{\β}}{2}\right)$

In the formula, I _tBe the effective irradiation intensity of photovoltaic panel, I _bBe direct projection irradiation intensity, I _dBe scattering irradiation intensity, I _hBe the total irradiation intensity of surface level, β is the photovoltaic panel inclination angle, θ _iBe solar incident angle, ρ is a reflection coefficient.Calculate the effective irradiation intensity I that to know photovoltaic panel through known data _t

In step b and the steps d, temperature is modified to the effective temperature of photovoltaic panel: T=T through following formula _Air+ KS

In the formula, T is the effective temperature of photovoltaic panel, T _AirBe environment temperature, S is an intensity of illumination, and K is a temperature coefficient.General value is 0.03 (℃ m ²/ w).

The input data of BP neural network also comprise time data, said time data comprises month, day and the time.Because between the output power of time data and photovoltaic plant bigger relation is arranged also, so with the input data of time data as the BP neural network, and be associated with the output power of photovoltaic plant, accuracy for predicting can be improved.In addition, the meteorological element data are not limited only to irradiation intensity, temperature and wind speed, also can comprise the meteorological element data that other are associated with generated energy.

Following table 1 is carried out the predicated error statistical form of power prediction for Harbor photovoltaic plant and Village Vanguard photovoltaic plant use the inventive method.

Table 1

From table, can find out the accuracy height that predicts the outcome of the inventive method, satisfy the requirement that operation of power networks is used.

The error statistics explanation:

1, average absolute value error (nMAE)

$\mathrm{nMAE}=\frac{{\mathrm{\Σ}}_{i=1}^n|P_{\mathrm{Mi}}-P_{\mathrm{Pi}}|}{\mathrm{Cap}\·n}$

P _MiBe i real power constantly, P _PiBe i predicted power constantly, Be the mean value of all sample real powers, Cap is the photovoltaic plant total installation of generating capacity, and n is all number of samples.

This index reflects the error of the absolute value that predicts the outcome, and can reflect the situation of error to a certain extent.But the point that error is bigger is submerged when doing statistical average easily, can not reflect the king-sized extreme case of error.

2, root-mean-square error (nRMSE)

$\mathrm{nRMSE}=\frac{\sqrt{{\mathrm{\Σ}}_{i=1}^n{(P_{\mathrm{Mi}}-P_{\mathrm{Pi}})}^2}}{\mathrm{Cap}\·n}$

P _MiBe i real power constantly, P _PiBe i predicted power constantly, Be the mean value of all sample real powers, Cap is the photovoltaic plant total installation of generating capacity, and n is all number of samples.

This index does not have the corresponding physical meaning.Because be that quadratic sum is opened radical sign, amplified the influence of the bigger point of error.In the timing statistics scope; Most of point prediction values depart from actual value about 10%; But the predicted value that a small amount of point is arranged departs from actual value more than 50% (such situation is bigger to the scheduling influence), adopts this error to embody the influence of these errors point bigger than normal 15%～20% even bigger.

3, error is no more than 20% the shared ratio (pre20) of point

$\mathrm{pre}20=\frac{n_p}{n}$

n _PSurpass the number of 20% point for error, n is all number of samples

Embodied the probability of deviation in tolerance interval of predicted value and actual value.Here 20% is the parameter that can change.

The advantage dispatcher of this index can weigh the risk of the margin capacity of preparing for photovoltaic generation power according to this index.Can make the probability level of a plurality of scopes according to actual needs, for example error is no more than the ratio of 10% point, and error is no more than ratio of 30% point etc.

Above embodiment is merely exemplary embodiment of the present invention, is not used in restriction the present invention, and protection scope of the present invention is defined by the claims.Those skilled in the art can make various modifications or be equal to replacement the present invention in essence of the present invention and protection domain, this modification or be equal to replacement and also should be regarded as dropping in protection scope of the present invention.

Claims (4)

1. based on the photovoltaic power station power generation output power Forecasting Methodology of meteorological element, it is characterized in that, comprise the steps:

A. obtain the on-site historical record of the meteorological element data of irradiation intensity, temperature and wind speed that comprises of photovoltaic plant and reach the photovoltaic power station power generation output power relative with each record;

B. wherein, irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel, temperature is modified to the effective temperature of photovoltaic panel, thereby generates revised meteorological element data;

C. revised meteorological element data are imported the BP neural network as the input data, photovoltaic power station power generation output power that will be corresponding with the meteorological element data is trained the BP neural network as the input of BP neural network;

D. obtain the meteorological element data that comprise irradiation intensity, temperature and wind speed of photovoltaic plant location according to numerical weather forecast in the predicted time section; And irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel; Temperature is modified to the effective temperature of photovoltaic panel, thereby generates revised meteorological element data;

E. with the BP neural network of revised meteorological element data input after step c training of steps d gained, the data of BP neural network output are the generating output power of the photovoltaic plant of this predicted time section.

2. the photovoltaic power station power generation output power Forecasting Methodology based on meteorological element according to claim 1 is characterized in that, in said step b and the steps d, irradiation intensity is modified to the effective irradiation intensity of photovoltaic panel through following formula:

$I_t=I_b\cos {\mathrm{\θ}}_i+I_d\left(\frac{1+\cos \mathrm{\β}}{2}\right)+\mathrm{\ρ}{I}_h\left(\frac{1-\cos \mathrm{\β}}{2}\right)$

In the formula, I _tBe the effective irradiation intensity of photovoltaic panel, I _bBe direct projection irradiation intensity, I _dBe scattering irradiation intensity, I _hBe the total irradiation intensity of surface level, β is the photovoltaic panel inclination angle, θ _iBe solar incident angle, ρ is a reflection coefficient.

3. the photovoltaic power station power generation output power Forecasting Methodology based on meteorological element according to claim 1 is characterized in that, in said step b and the steps d, temperature is modified to the effective temperature of photovoltaic panel: T=T through following formula _Air+ KS

In the formula, T is the effective temperature of photovoltaic panel, T _AirBe environment temperature, S is an intensity of illumination, and K is a temperature coefficient.

4. the photovoltaic power station power generation output power Forecasting Methodology based on meteorological element according to claim 1 is characterized in that the input data of BP neural network also comprise time data, said time data comprises month, day and the time.