CN109767353A - A kind of photovoltaic power generation power prediction method based on probability-distribution function - Google Patents
A kind of photovoltaic power generation power prediction method based on probability-distribution function Download PDFInfo
- Publication number
- CN109767353A CN109767353A CN201910031330.XA CN201910031330A CN109767353A CN 109767353 A CN109767353 A CN 109767353A CN 201910031330 A CN201910031330 A CN 201910031330A CN 109767353 A CN109767353 A CN 109767353A
- Authority
- CN
- China
- Prior art keywords
- photovoltaic
- generation power
- probability
- value
- distribution function
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Photovoltaic Devices (AREA)
Abstract
The present invention relates to a kind of photovoltaic power generation power prediction methods based on probability-distribution function, comprising the following steps: step 1: collecting the irradiance data and corresponding generated output data of photovoltaic generating system to be predicted；Step 2: the value range of irradiation level is divided into M section；In each section, it is fitted to obtain the corresponding photovoltaic generation power probability density function in the section and Cumulative Distribution Function using general distribution fitting method；Step 3: for any known irradiation level, finding the section corresponding to it and photovoltaic generation power probability distributing density function, obtain several corresponding photovoltaic generation power values of known irradiation level, each photovoltaic generation power value constitutes photovoltaic power output scene collection；Step 4: to photovoltaic power output scene concentrate each photovoltaic generation power value averaged as the corresponding photovoltaic power generation power prediction value of known irradiation level.The present invention can it is quick and it is high-precision photovoltaic generation power is predicted, practical application value with higher.
Description
Technical field
The invention belongs to field of photovoltaic power generation, and in particular to the method that a kind of pair of photovoltaic generating system carries out power prediction.
Background technique
Photovoltaic power generation have non-pollutant discharge, without fuel consumption, capacity scale is unrestricted, application form is flexible, pacify
The advantages that complete reliable, maintenance is simple, therefore there is vast potential for future development.Due to the support of countries in the world policy, light in recent years
The development for lying prostrate power generation is very fast, has been carried out fairly large application in the world.It is newest according to German solar energy association
The statistical data of publication shows that the global newly-increased installation 70GW of photovoltaic in 2016 increased about 30%, so far, global light than 2015
Volt installation total amount reaches 300GW.
Photovoltaic generating system is the electricity generation system that electric energy is converted solar energy into using photovoltaic effect.Its is most basic
Composition includes photovoltaic module, battery, controller, inverter and support system.In general, photovoltaic generating system also will be installed it is small
Type Meteorological Device and some monitoring systems are to measure and record meteorological and system performance information.Photovoltaic generating system forms substantially
As shown in Fig. 1.
The generated output of photovoltaic generating system is predicted, is using an important work during photovoltaic generating system
Make, and during the prediction, it is necessary to need to use photovoltaic generation power model, i.e., in one group of given weather prognosis information bar
Under part, photovoltaic generation power model can be used for predicting photovoltaic power generation system output power, be photovoltaic power generation power prediction can not or
Scarce link.In addition to this, before purchase of equipment, photovoltaic generation power model can be used for comparing the anticipated output of particular system
It is designed with other possible systems.Finally, photovoltaic generation power model can also be used to determine system whether run on schedule, make be
System operator can determine maintenance plan.Therefore, research photovoltaic generation power model has certain practical guided significance, photovoltaic
The determination of generation model directly influences the effect generated after its application.
Currently, the modeling method in relation to photovoltaic generation power model mainly has physical model and statistical model two both at home and abroad
Class.Physical modeling be conceived to portray from solar energy to electrical conversion process each energy conversion device (photovoltaic cell, controller,
Inverter etc.) and operation control system module mathematical relationship, the validity of this method depends on to research object inside structure
At and its institute's follow regularity assurance degree and model parameter precision.The most important link for influencing physical model precision is light
Battery model is lied prostrate, most common method is to establish the photovoltaic cell equivalent-circuit model based on diode.Earliest model is from short circuit
Circuit, open-circuit voltage and Diode Ideality Factor are started with, and are proposed by a Line independent current source and a parallel diode
The three parameter model of composition.This method is succinctly easy to operate, but precision is lower.In turn, document proposes to pass through the side of series resistance Rs
Formula improves three parameter model, which is widely referred to as Rs model, and Rs model is so far using most extensively
Model, but its computational accuracy substantially reduces in the higher situation of temperature.In this regard, pertinent literature is on the basis of Rs model,
An additional parallel resistance Rp is increased, Rp model is extended to.Although the model is promoted in precision, with
Being continuously increased for model parameter, parameter tuning is further difficult, computationally intensive.Therefore it on the basis of Rp five-parameter model, proposes
Simplified photovoltaic cell engineering model, four basic parameters which only needs photovoltaic cell manufacturer to provide are i.e.
Photovoltaic cell power producing characteristics can be obtained.Engineering model reduces the calculation amount of model to a certain extent, but practice have shown that, by
It is influenced in by photovoltaic generating system efficiency, precision of the physical model in low irradiance is lower.
Different from physical model, statistical model more pays close attention to the mathematical statistics between the input and output of photovoltaic generating system
Rule, and desalinate the analysis of the characteristic and internal each influence factor of internal each module.It is based on actual operating data and goes out to photovoltaic
Force characteristic is fitted, and then reflects the Nonlinear Mapping between input and output.Common statistical model modeling method has mind
Through network, associated data etc.: document assesses the performance of photovoltaic array using neural network, further refreshing using returning
The power output model of photovoltaic generating system photovoltaic generating system is established through network, and this model and physical model modeling method are done
Comparison, it is indicated that recurrent neural network model is better than physical model in computational accuracy.Although neural network precision is higher,
Be its numerous hidden neuron and best hidden neuron number and best synapse weight determine need complicated algorithm and
A large amount of research work.
It can be seen that based on defect existing for existing photovoltaic generation power model, so that existing photovoltaic power generation function
Rate prediction has that precision is lower or algorithm is complicated, heavy workload.
Summary of the invention
It is higher and relatively simple, easy to implement based on probability-distribution function the object of the present invention is to provide a kind of precision
Photovoltaic power generation power prediction method.
In order to achieve the above objectives, the technical solution adopted by the present invention is that:
A kind of photovoltaic power generation power prediction method based on probability-distribution function, comprising the following steps:
Step 1: collecting the irradiance data and corresponding power generation of photovoltaic generating system to be predicted in the historical data
Power data, line number of going forward side by side Data preprocess；
Step 2: the value range of irradiation level is divided into M section；It is quasi- using general distribution in each section
Conjunction method is fitted to obtain the corresponding photovoltaic generation power probability density function in the section and Cumulative Distribution Function, each area
Between corresponding photovoltaic generation power probability distributing density function constitute general distribution probability density function library；
Step 3: for any known irradiation level, finding the section corresponding to it and photovoltaic generation power probability point
Cloth density function carries out inverse transformation sampling to the photovoltaic generation power probability distributing density function and obtains the known irradiation
Several corresponding photovoltaic generation power values are spent, each photovoltaic generation power value constitutes photovoltaic power output scene collection；
Step 4: described in each photovoltaic generation power value averaged conduct concentrated to photovoltaic power output scene
The corresponding photovoltaic power generation power prediction value of known irradiation level.
In the step 1, the data prediction includes deletion error data, rejects hash and historical data mark
Change.
In the step 2,0~1p.u. of value range of irradiation level is divided into the M sections, each section
Width be the irradiation level value range 1/M.
In the step 3,0~1p.u. of value range of irradiation level is divided into the width not equal M section, is made
Corresponding each generated output data form unimodal state in each section.
In the step 2, the corresponding probability density function expression formula of the general distribution fitting method that uses forCumulative Distribution Function expression formula is F (x)=(1+e-α(x-γ))-β, wherein α, β and γ are respectively shape
Shape parameter, and meet α > 0, β > 0 ,-∞ < γ <+∞.
In the step 3, the method for inverse transformation sampling are as follows: introduce stochastic variable Zt, utilizeSeek the stochastic variable ZtCorresponding standard normal distribution function value Φ (Zt), recycle Pt
=Fl-1(Φ(Zt)) seek standard normal distribution function value Φ (Zt) corresponding photovoltaic generation power value, wherein Fl-1(U) it is
Cumulative Distribution Function FlInverse function.
In the step 4, each photovoltaic generation power value concentrated to photovoltaic power output scene is sought being weighted and averaged
Value is as the corresponding photovoltaic power generation power prediction value of the known irradiation level.
In the step 2, variable temperature correction facotor is introduced to the photovoltaic generation power probability density function and is tired out
Meter distribution function is corrected.
The expression formula of the temperature correction facotor is α=(Pt cal-Pt actual)/(Tt-Tt-1), wherein α is temperature correction system
Number, t is moment, Pt calFor t moment photovoltaic power generation power prediction value, Pt actualFor t moment photovoltaic power output actual value, TtWhen for t
Carve temperature, Tt-1For t-1 moment temperature.
Each section, to minimize photovoltaic generation power root-mean-square error as target, when obtaining multiple and different
It carves, the temperature correction basic coefficients under the conditions of different irradiation level, taking the average value of each temperature correction basic coefficients to be used as should
The temperature correction facotor in the section.
Due to the above technical solutions, the present invention has the following advantages over the prior art: the present invention can be quick
And it is high-precision photovoltaic generation power is predicted, practical application value with higher.
Detailed description of the invention
Attached drawing 1 is the basic composition schematic diagram of photovoltaic generating system.
Attached drawing 2 is the fitting comparison diagram of general distribution, Gaussian Profile and Weibull distribution.
Attached drawing 3 is inverse transformation sampling flow chart.
Attached drawing 4 is the flow chart of the photovoltaic power generation power prediction method based on probability-distribution function.
Attached drawing 5 is the dispersed big and Double-peak Phenomenon schematic diagram of historical data.
Attached drawing 6 is fitting effect schematic diagram after uneven branch mailbox improves.
Attached drawing 7 is each the model calculation schematic diagram.
Specific embodiment
The invention will be further described for embodiment shown in reference to the accompanying drawing.
A kind of embodiment one: photovoltaic power generation power prediction method based on probability-distribution function, comprising the following steps:
Step 1: collecting the irradiance data and corresponding power generation of photovoltaic generating system to be predicted in the historical data
Power data, line number of going forward side by side Data preprocess.
It is more than the information for including due to obtained initial data and miscellaneous, it is therefore desirable to first historical data obtained to be carried out pre-
Processing.Data prediction includes deletion error data, rejects hash and the change of historical data mark, specific as follows:
A) data needed for extracting.By required data, i.e. irradiance data, temperature data and photovoltaic goes out force data and (sends out
Electrical power data) it is extracted from historical data.
B) deletion error data.Firstly, irradiation level is deleted with the obvious not corresponding wrong data of photovoltaic power output；Secondly,
Due to the measurement error near sunrise moment and sunset moment, certain irradiation level are negative value, these data are also wrong data, are answered
Work as deletion.
C) hash is rejected.Before sunrise and post sunset irradiation level is zero, and photovoltaic power output is also zero, is gone out for establishing photovoltaic
For power model, these data are hash, should be rejected.
D) historical data mark is changed.After rejecting wrong data and hash, study hereinafter for convenience, most with irradiation level
Big value and photovoltaic maximum output are base value, convert per unit value for all historical datas.
Step 2: 0~1p.u. of value range of irradiation level being divided into M section, M is positive integer, and each section is corresponding
For an irradiation level grade, it is referred to as " case " in text, therefore irradiation level demarcation interval is also referred to as " branch mailbox ".Common mode
It is that the value range of irradiation level is divided into M section, the width in each section is the 1/M of the value range of irradiation level.And base
In the value range 0~1 of irradiation level, therefore the value range of the corresponding irradiation level in each section is respectively [0,1/M], [1/M, 2/
M], [2/M, 3/M] ..., [(M-1)/M, 1].The value range of m (m=1,2 ..., M) corresponding irradiation level of grade be [(m-1)/
M,m/M]。
The practical power output original data set of the corresponding photovoltaic of m-th of irradiation level grade: each in original data set
There are an irradiance value and a practical power generating value of photovoltaic in a period.For each irradiance value for belonging to m grades, have
One corresponding practical power generating value of photovoltaic, the set that all these practical power generating values are constituted are referred to as " m-th of irradiation level grade
The corresponding practical power output original data set of photovoltaic ".
It is noted that M is the customized positive integer of user, available historical data is more, and M value is bigger, on the contrary
It is smaller.On the other hand, if historical data total amount is fixed, M value is bigger, the corresponding practical power output initial data of photovoltaic of each case
Sample size is fewer in set, otherwise more.The practical power output original data set of photovoltaic corresponding for each case, if the collection
The sample size of conjunction is very few, will reduce the confidence level of statistical result, this just loses the meaning of statistics.Therefore, hereinafter for guarantee
The effect of middle statistical analysis should make sample size in the practical power output original data set of the corresponding photovoltaic of each case enough.
Since certain distribution character is not presented in overall data, if branch mailbox is not directly to all history irradiance datas
Go out force data with photovoltaic to be fitted, will lead to that fitting effect is poor, and fitting precision is low, it is therefore desirable to which branch mailbox is carried out to historical data
Processing.
After completion data branch mailbox obtains the practical power output original data set of the corresponding photovoltaic of each case, need to count it
According to fitting, that is, establish probability distribution of the practical power output of photovoltaic in each case.In the present embodiment, in each section, using logical
It is fitted to obtain the corresponding photovoltaic generation power probability density function in the section and Cumulative Distribution Function (probability with distribution fitting method
Distributed model), so that the corresponding photovoltaic generation power probability distributing density function in each section constitutes general distribution probability density letter
Number library.The corresponding probability density function expression formula of general distribution fitting method used are as follows:
Cumulative Distribution Function expression formula are as follows:
F (x)=(1+e-α(x-γ))-β (2)
Wherein α, β and γ are respectively form parameter, and are met:
α > 0, β > 0 ,-∞ < γ <+∞ (3)
General distribution has following advantageous property:
A) there are three form parameters for general distribution tool, and by adjusting three form parameters, the PDF and CDF of general distribution are bent
Line can be deformed flexibly, so that it is farthest approached the practical probability distribution contributed of photovoltaic, therefore general distribution can be more smart
The practical probability distribution contributed of photovoltaic under the conditions of the true any irradiation level of characterization.
B) CDF of general distribution and its inverse function all have the closure expression formula of parsing, are gone out with general distribution characterization photovoltaic
The probability distribution of power can simplify the calculating process of model.
The present invention chooses two kinds of typical subcases (m=7, m=11) and is fitted Contrast on effect analysis, and error of fitting is such as
Shown in table 1, fitting effect is as shown in Figure 2.
The general distribution of table 1, Gaussian Profile and the comparison of Weibull Distribution error
As shown in Table 1, general distribution is minimum for the error of fitting of actual distribution, and fitting precision highest, fitting effect is most
It is good.
As shown in Figure 2, when photovoltaic power output is less than normal, actual distribution can be to left avertence, and thick tail phenomenon occurs；Work as photovoltaic
When contributing bigger than normal, actual distribution can to right avertence, and occur thickness appear for the first time as.Gaussian Profile is symmetrical, therefore cannot be characterized
The off-axis characteristic of actual distribution；Although Weibull distribution has off-axis characteristic, fitting effect is bad, as shown in Fig. 2 (a),
Weibull distribution is unable to the first and thick tail phenomenon of thickness of accurate Characterization actual distribution；In contrast, since general distribution has three ginsengs
Number, thus can be more accurate characterization actual distribution off-axis characteristic and thick first and thick tail phenomenon, and general be distributed pair
It is also more accurate in the fitting of actual distribution peak value.
Step 3: for any known irradiation level, finding section corresponding to it and photovoltaic generation power probability distribution is close
Function is spent, it is corresponding several to obtain known irradiation level to the progress inverse transformation sampling of photovoltaic generation power probability distributing density function
A photovoltaic generation power value, each photovoltaic generation power value constitute photovoltaic power output scene collection.
Specifically, for any known irradiation intensity, judging which it belongs to after carrying out branch mailbox fitting to historical data
A case simultaneously finds corresponding general distribution function, carries out inverse transformation sampling to general distribution by generating a large amount of random numbers,
It obtains photovoltaic power output probable value and obtains scene collection.The basic skills of inverse transformation sampling is as follows:
Assuming that we need to obedience Pr(Pt≤ p)=Fl(p) a certain stochastic variable is sampled, i.e., from stochastic variable
PtCDF carry out random sampling, inverse transformation sampling mode it is as follows:
Pt=Fl -1(U), [0,1] (4) U~Unif
In formula, Fl -1It (U) is accumulated probability distribution FlInverse function；In the present invention, which is distributed FlIt is general
Distribution；Unif [0,1] indicates being uniformly distributed between [0,1].
Introduce the stochastic variable Z of an obedience standardized normal distributiont, standard deviation is 1, is contemplated to be 0.If a large amount of generate clothes
From the random number of standardized normal distribution, the standard normal distribution function value set of these random numbers is equal between obedience [0,1]
Even distribution.So, being uniformly distributed U and can use standard normal distribution function value Φ (Z in formula (4)t) substitution.Known random change
Measure ZtRandom number when, to stochastic variable PtIt can be sampled using following formula:
It utilizes
Seek variable ZtCorresponding standard normal distribution function value Φ (Zt), it recycles
Pt=Fl -1(Φ(Zt)) (6)
Seek standard normal distribution function value Φ (Zt) corresponding photovoltaic generation power value, wherein Fl -1It (U) is accumulative point
Cloth function FlInverse function.
Inverse transformation sampling process can be indicated by Fig. 3.
Step 4: each photovoltaic generation power value concentrated to photovoltaic power output scene seeks weighted average as known irradiation
Spend corresponding photovoltaic power generation power prediction value (desired value of photovoltaic power output).
The flow chart of the photovoltaic power generation power prediction method based on probability-distribution function can be indicated by Fig. 4 above.
Improvement to the photovoltaic power generation power prediction method above based on probability-distribution function:
1, uneven branch mailbox
It is found during carrying out branch mailbox fitting to historical data, if the history using uniform branch mailbox, in certain casees
Data are although more, but data dispersibility is larger or even bimodal phenomenon occurs, as shown in Figure 5.These phenomenons all can compared with
Reduce fitting precision in big degree, it is therefore desirable to which secondary branch mailbox, i.e., uneven branch mailbox are carried out to these casees.In other words, in step
In rapid 3,0~1p.u. of value range of irradiation level is divided into width not M equal section, is made corresponding each in each section
A generated output data form unimodal state.Fitting effect is as shown in Figure 6 after being improved using uneven branch mailbox.It can by Fig. 5 and Fig. 6
Know, after being improved using uneven branch mailbox, historical data dispersibility greatly reduces and Double-peak Phenomenon, therefore fitting effect no longer occurs
It is greatly improved with fitting precision.It is as shown in table 2 to improve front and back error of fitting:
Uneven branch mailbox error of fitting comparison after the uniform branch mailbox of table 2 and improvement
2, variable coefficient temperature correction
Above-mentioned model only considered the influence that irradiation level contributes to photovoltaic, and influence most important two factors of photovoltaic power output
It is irradiation level and component temperature, it is therefore desirable to introduce temperature correction link, existing method is corrected using steady temperature at present
Coefficient is corrected model.The present invention proposes to introduce variable temperature correction facotor to photovoltaic generation power probability density function
It is corrected with Cumulative Distribution Function, under reflecting history synchronization condition of different temperatures, the calculating error of model is corrected with this
Photovoltaic power output model, the expression formula of temperature correction facotor are
α=(Pt cal-Pt actual)/(Tt-Tt-1) (7)
In above formula, α is temperature correction facotor, and t is moment, Pt calModel predication value is contributed (before utilizing for t moment photovoltaic
State the photovoltaic power generation power prediction value that the photovoltaic power generation power prediction method based on probability-distribution function obtains), Pt actualWhen for t
Carve photovoltaic power output actual value, TtFor t moment temperature, Tt-1For t-1 moment temperature.
Branch mailbox thought is equally taken, each section is obtained using minimizing photovoltaic generation power root-mean-square error as target
Temperature correction basic coefficients under the conditions of multiple and different moment, different irradiation level, take being averaged for each temperature correction basic coefficients
It is worth the temperature correction facotor as the section, i.e., the temperature correction facotor in each case is different, to improve model correction
Precision, therefore referred to as temperature correction variable coefficient.
Specific correcting process is as follows:
A) photovoltaic power output model predication value and photovoltaic power output actual value are divided by M section according to irradiation level per unit value, often
There are several photovoltaics power output model predication value and the practical Value Data pair of photovoltaic power output in a section, and finds its temperature for corresponding to the moment
Value.
B) adjacent moment photovoltaic as caused by temperature change power output model predication value in each section is calculated according to formula (7)
With the difference of photovoltaic power output actual value, the temperature coefficient of adjacent moment in each section is thus calculated, temperature system in the section is taken
Number average value is the temperature correction facotor α in the sectionavg。
C) for each prediction time, judge that the moment irradiation level predicted value is in m-th of section, and utilize the section
Probability Distribution Model acquire photovoltaic power output model predication value, further according to adjacent moment temperature difference and formula (8) to its predicted value into
Trip temperature correction:
In formula (8), t is the moment,For photovoltaic power generation output forecasting value after t+1 moment temperature correction,For m-th of section
Temperature correction facotor (i.e. each temperature coefficient average value in the section),The probability distribution in the section is utilized for the t+1 moment
The resulting photovoltaic power generation output forecasting value of model, TtFor t moment temperature, Tt+1For t+1 moment temperature,For correcting value.
Numerical results and analysis
Historical data of the present invention derive from roof photovoltaic power station, photovoltaic plant installed capacity 9.066MW, by
Two kinds of photovoltaic modulies, 18 inverter compositions, two kinds of photovoltaic module basic parameters are as shown in table 3:
3 two kinds of photovoltaic module basic parameters of table
Engineering physical model, probabilistic model (uniform branch mailbox, uneven point based on general distribution is respectively adopted in the present invention
Case, uneven branch mailbox and variable coefficient temperature correction) photovoltaic power output is calculated, each the model calculation is as shown in fig. 7, each
It is as shown in table 4 that model calculates error:
Each model of table 4 calculates error
By Fig. 7 and table 4 it is found that when irradiation level is lower, engineering is larger with physical model calculating error, after sunrise and day
Model calculation value and actual value deviation are larger in several hours before falling, therefore model overall calculation error is larger, and work as irradiation level
When higher, engineering is higher with physical model calculating precision.Compared to engineering physical model, the probabilistic model based on general distribution
Overall calculation precision is higher, and overall calculation error is smaller: uniform branch mailbox substantially increases computational accuracy when low irradiance, however
Computational accuracy decreases near photovoltaic power output peak value, and overall calculation error greatly reduces；On the basis of being uniformly distributed, no
It is uniformly distributed computational accuracy when further improving low irradiance, however the calculating error near photovoltaic power output peak value is not
It is improved, overall calculation error slightly reduces；On the basis of uneven branch mailbox, uneven branch mailbox and temperature correction model are not
Computational accuracy when low irradiance is only further improved, and the calculating error near photovoltaic power output peak value is also changed
Into overall calculation error greatly reduces.The calculating time of all models all within 0.3s, meets the requirement of rapidity.
Engineering is the theoretical value of photovoltaic power output and the product of photovoltaic generating system efficiency with physical model calculating value, due to ash
Dirt, rainwater block, component is connected mismatch, the power loss of inverter, DC communication part cable power loss, transformer function
The factors such as the precision of rate loss and tracking system, the efficiency of photovoltaic generating system usually only 0.8, so that
Engineering physical model calculating value=photovoltaic power output theoretical value * 0.8
However due to the fact that, in low irradiance, the actual efficiency of photovoltaic generating system is often much smaller than 0.8:
A) loss of photovoltaic module, inverter, cable etc. is not linear change, and when low irradiance is lost larger；
B) when radiating very low, there is group a string brownout, the case where part inverter can not start；
C) inverter accounts for hair in low irradiation in the presence of certain from power consumption condition (from power consumption with load variations very little)
Electricity specific gravity is higher, and system effectiveness is caused to reduce.
Therefore in low irradiance, photovoltaic power output actual value is much smaller than engineering physical model calculating value.And it is based on general point
What the probabilistic model of cloth was established is irradiation level, temperature and the practical non-linear relation contributed of photovoltaic, during establishing model,
It include that in a model, therefore whole precision of prediction is high by the efficiency of photovoltaic generating system.
In conclusion the present invention is fitted historical data using general distribution, establish in different irradiation level sections
The probability distribution of photovoltaic generation power, and uneven branch mailbox data fitting method is proposed, improve the precision of data fitting.This
Invention proposes variable coefficient temperature correction model and is corrected to photovoltaic power output, to minimize photovoltaic generation power root mean square mistake
Difference is target, has obtained the temperature correction facotor under the conditions of different moments, different irradiation level and has further improved the precision of model.
In the present invention, the probability distribution of a kind of characterization photovoltaic generation power and irradiation level and temperature map relationship is first proposed
Modeling method, the model do not need the detail parameters of each physical module, are not related to complicated derivation algorithm, therefore be based on the mould yet
When type carries out photovoltaic power generation power prediction, calculating speed is very fast, and compared to traditional engineering physical model, computational accuracy is significantly
It is promoted；Then propose uneven branch mailbox and variable coefficient temperature correction improve model, calculate column the result shows that, after improvement
Model, which is appointed, so has rapidity, and computational accuracy further increases.It, should in the case where known one group of meteorological condition predicted value
Model can be used to the accurate predicted value for calculating photovoltaic power output under this condition, and the raising of the model accuracy is to realize photovoltaic power generation function
The important link of rate Accurate Prediction, therefore there is very high practical application value.
The above embodiments merely illustrate the technical concept and features of the present invention, and its object is to allow person skilled in the art
Scholar cans understand the content of the present invention and implement it accordingly, and it is not intended to limit the scope of the present invention.It is all according to the present invention
Equivalent change or modification made by Spirit Essence, should be covered by the protection scope of the present invention.
Claims (10)
1. a kind of photovoltaic power generation power prediction method based on probability-distribution function, it is characterised in that: the photovoltaic generation power
Prediction technique the following steps are included:
Step 1: collect in the historical data photovoltaic generating system to be predicted irradiance data and corresponding generated output
Data, line number of going forward side by side Data preprocess；
Step 2: the value range of irradiation level is divided into M section；In each section, using general fitting of distribution side
Method is fitted to obtain the corresponding photovoltaic generation power probability density function in the section and Cumulative Distribution Function, each section pair
The photovoltaic generation power probability distributing density function answered constitutes general distribution probability density function library；
Step 3: for any known irradiation level, finding the section corresponding to it and photovoltaic generation power probability distribution is close
Function is spent, inverse transformation sampling is carried out to the photovoltaic generation power probability distributing density function and obtains the known irradiation level pair
Several photovoltaic generation power values answered, each photovoltaic generation power value constitute photovoltaic power output scene collection；
Step 4: to the photovoltaic power output scene concentrate each photovoltaic generation power value averaged as described known
The corresponding photovoltaic power generation power prediction value of irradiation level.
2. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 1, the data prediction includes deletion error data, rejects hash and the change of historical data mark.
3. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 2,0~1p.u. of value range of irradiation level is divided into the M sections, the width in each section
For the 1/M of the value range of the irradiation level.
4. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 3,0~1p.u. of value range of irradiation level is divided into the width not equal M section, makes each institute
It states corresponding each generated output data in section and forms unimodal state.
5. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 2, the corresponding probability density function expression formula of the general distribution fitting method that uses forCumulative Distribution Function expression formula is F (x)=(1+e-α(x-γ))-β, wherein α, β and γ are respectively shape
Shape parameter, and meet α > 0, β > 0 ,-∞ < γ <+∞.
6. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 3, the method for inverse transformation sampling are as follows: introduce stochastic variable Zt, utilizeIt seeks
The stochastic variable ZtCorresponding standard normal distribution function value Φ (Zt), recycle Pt=Fl -1(Φ(Zt)) seeking the standard just
State distribution function value Φ (Zt) corresponding photovoltaic generation power value, whereinIt is Cumulative Distribution Function FlInverse function.
7. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 1, feature exist
In: in the step 4, each photovoltaic generation power value concentrated to photovoltaic power output scene seeks weighted average conduct
The corresponding photovoltaic power generation power prediction value of the known irradiation level.
8. a kind of photovoltaic power generation power prediction side based on probability-distribution function according to any one of claim 1 to 7
Method, it is characterised in that: in the step 2, introduce variable temperature correction facotor to the photovoltaic generation power probability density letter
Several and Cumulative Distribution Function is corrected.
9. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 8, feature exist
In: the expression formula of the temperature correction facotor is α=(Pt cal-Pt actual)/(Tt-Tt-1), wherein α is temperature correction facotor, t
For moment, Pt calFor t moment photovoltaic power generation power prediction value, Pt actualFor t moment photovoltaic power output actual value, TtFor t moment temperature
Degree, Tt-1For t-1 moment temperature.
10. a kind of photovoltaic power generation power prediction method based on probability-distribution function according to claim 9, feature exist
In: each section, to minimize photovoltaic generation power root-mean-square error as target, obtain multiple and different moment, no
With the temperature correction basic coefficients under the conditions of irradiation level, take the average value of each temperature correction basic coefficients as area described in this
Between temperature correction facotor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031330.XA CN109767353B (en) | 2019-01-14 | 2019-01-14 | Photovoltaic power generation power prediction method based on probability distribution function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031330.XA CN109767353B (en) | 2019-01-14 | 2019-01-14 | Photovoltaic power generation power prediction method based on probability distribution function |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109767353A true CN109767353A (en) | 2019-05-17 |
| CN109767353B CN109767353B (en) | 2020-12-18 |
Family
ID=66452777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910031330.XA Active CN109767353B (en) | 2019-01-14 | 2019-01-14 | Photovoltaic power generation power prediction method based on probability distribution function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109767353B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111476411A (en) * | 2020-04-01 | 2020-07-31 | 阳光电源（上海）有限公司 | Photovoltaic power station generated energy calculation method and server |
| CN111861259A (en) * | 2020-07-30 | 2020-10-30 | 中国大唐集团科学技术研究院有限公司华东电力试验研究院 | Load modeling method, system and storage medium considering time sequence |
| CN113570116A (en) * | 2021-07-02 | 2021-10-29 | 福建时代星云科技有限公司 | Method for estimating heating power of storage and charging station and terminal |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103208029A (en) * | 2013-03-11 | 2013-07-17 | 中国电力科学研究院 | Super-short-term power prediction method based on clearance model for photovoltaic power station |
| CN104657791A (en) * | 2015-02-28 | 2015-05-27 | 武汉大学 | Wind power plant group wind speed distribution prediction method based on correlation analysis |
| CN105023070A (en) * | 2015-08-12 | 2015-11-04 | 河海大学常州校区 | Output power prediction method of photovoltaic system |
| CN105490310A (en) * | 2016-01-25 | 2016-04-13 | 武汉大学 | Real-time dynamic economical scheduling method for generally-distributed electric power system containing wind power |
| CN105846425A (en) * | 2016-04-08 | 2016-08-10 | 江苏省电力试验研究院有限公司 | Economic dispatching method based on general wind power forecasting error model |
| CN105930671A (en) * | 2016-04-29 | 2016-09-07 | 武汉大学 | Improved versatile distribution and versatile mixture distribution models characterizing wind power probability distribution |
| US20170228834A1 (en) * | 2011-07-25 | 2017-08-10 | Clean Power Research, L.L.C. | Generating A Risk-Adjusted Probabilistic Forecast Of Renewable Power Production For A Fleet With The Aid Of A Digital Computer |
-
2019
- 2019-01-14 CN CN201910031330.XA patent/CN109767353B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170228834A1 (en) * | 2011-07-25 | 2017-08-10 | Clean Power Research, L.L.C. | Generating A Risk-Adjusted Probabilistic Forecast Of Renewable Power Production For A Fleet With The Aid Of A Digital Computer |
| CN103208029A (en) * | 2013-03-11 | 2013-07-17 | 中国电力科学研究院 | Super-short-term power prediction method based on clearance model for photovoltaic power station |
| CN104657791A (en) * | 2015-02-28 | 2015-05-27 | 武汉大学 | Wind power plant group wind speed distribution prediction method based on correlation analysis |
| CN105023070A (en) * | 2015-08-12 | 2015-11-04 | 河海大学常州校区 | Output power prediction method of photovoltaic system |
| CN105490310A (en) * | 2016-01-25 | 2016-04-13 | 武汉大学 | Real-time dynamic economical scheduling method for generally-distributed electric power system containing wind power |
| CN105846425A (en) * | 2016-04-08 | 2016-08-10 | 江苏省电力试验研究院有限公司 | Economic dispatching method based on general wind power forecasting error model |
| CN105930671A (en) * | 2016-04-29 | 2016-09-07 | 武汉大学 | Improved versatile distribution and versatile mixture distribution models characterizing wind power probability distribution |
Non-Patent Citations (2)
| Title |
|---|
| 张纯杰 等: "光伏电池的建模综述", 《电源技术》 * |
| 曹慧秋 等: "基于通用分布的风电功率出力动态场景生成方法", 《电测与仪表》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111476411A (en) * | 2020-04-01 | 2020-07-31 | 阳光电源（上海）有限公司 | Photovoltaic power station generated energy calculation method and server |
| CN111476411B (en) * | 2020-04-01 | 2023-10-31 | 阳光电源（上海）有限公司 | Photovoltaic power station generating capacity calculation method and server |
| CN111861259A (en) * | 2020-07-30 | 2020-10-30 | 中国大唐集团科学技术研究院有限公司华东电力试验研究院 | Load modeling method, system and storage medium considering time sequence |
| CN111861259B (en) * | 2020-07-30 | 2022-07-15 | 中国大唐集团科学技术研究院有限公司华东电力试验研究院 | Load modeling method, system and storage medium considering time sequence |
| CN113570116A (en) * | 2021-07-02 | 2021-10-29 | 福建时代星云科技有限公司 | Method for estimating heating power of storage and charging station and terminal |
| CN113570116B (en) * | 2021-07-02 | 2022-02-18 | 福建时代星云科技有限公司 | Method for estimating heating power of storage and charging station and terminal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109767353B (en) | 2020-12-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107832905B (en) | Power distribution network planning method suitable for distributed power generation and energy storage station development | |
| CN109858673A (en) | A kind of photovoltaic generating system power forecasting method | |
| CN106532778B (en) | Method for calculating maximum access capacity of distributed photovoltaic grid connection | |
| CN107194495B (en) | Photovoltaic power longitudinal prediction method based on historical data mining | |
| CN105184678A (en) | Method for constructing photovoltaic power station generation capacity short-term prediction model based on multiple neural network combinational algorithms | |
| CN105260800A (en) | Photovoltaic module temperature prediction method and device | |
| CN110334870B (en) | Photovoltaic power station short-term power prediction method based on gated cyclic unit network | |
| CN108614612B (en) | Method and system for tracking maximum power of solar photovoltaic cell | |
| CN109086928A (en) | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model | |
| CN110380444B (en) | Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula | |
| CN106529719A (en) | Method of predicting wind power of wind speed fusion based on particle swarm optimization algorithm | |
| CN109995090B (en) | Multi-energy complementation method considering user demand characteristics and energy utilization cost | |
| CN109767353A (en) | A kind of photovoltaic power generation power prediction method based on probability-distribution function | |
| CN109636054A (en) | Solar energy power generating amount prediction technique based on classification and error combination prediction | |
| CN107145707A (en) | It is a kind of to count and photovoltaic is exerted oneself the power distribution network transformer planing method of uncertain and overall life cycle cost | |
| CN105305488B (en) | A kind of evaluation method for considering new-energy grid-connected and power transmission network utilization rate being influenced | |
| CN114792156A (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| Kasprzyk et al. | Minimisation of the LCOE for the hybrid power supply system with the lead-acid battery | |
| CN111784543A (en) | System model for predicting short-term photovoltaic power generation system power generation based on hidden Markov model | |
| CN115441447B (en) | New energy power generation power prediction method | |
| CN116629416A (en) | Photovoltaic power station power prediction method and device | |
| CN117485177A (en) | Intelligent energy-saving control system and method for charging pile | |
| CN108694475B (en) | Short-time-scale photovoltaic cell power generation capacity prediction method based on hybrid model | |
| Liu et al. | A weight-varying ensemble method for short-term forecasting PV power output | |
| Ni et al. | A review of line loss analysis of the low-voltage distribution system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |