AU2021104436A4 - Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day - Google Patents
Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day Download PDFInfo
- Publication number
- AU2021104436A4 AU2021104436A4 AU2021104436A AU2021104436A AU2021104436A4 AU 2021104436 A4 AU2021104436 A4 AU 2021104436A4 AU 2021104436 A AU2021104436 A AU 2021104436A AU 2021104436 A AU2021104436 A AU 2021104436A AU 2021104436 A4 AU2021104436 A4 AU 2021104436A4
- Authority
- AU
- Australia
- Prior art keywords
- predicted
- power generation
- day
- historical
- generation capacity
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/048—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators using a predictor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F21—LIGHTING
- F21S—NON-PORTABLE LIGHTING DEVICES; SYSTEMS THEREOF; VEHICLE LIGHTING DEVICES SPECIALLY ADAPTED FOR VEHICLE EXTERIORS
- F21S8/00—Lighting devices intended for fixed installation
- F21S8/08—Lighting devices intended for fixed installation with a standard
- F21S8/085—Lighting devices intended for fixed installation with a standard of high-built type, e.g. street light
- F21S8/086—Lighting devices intended for fixed installation with a standard of high-built type, e.g. street light with lighting device attached sideways of the standard, e.g. for roads and highways
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F21—LIGHTING
- F21S—NON-PORTABLE LIGHTING DEVICES; SYSTEMS THEREOF; VEHICLE LIGHTING DEVICES SPECIALLY ADAPTED FOR VEHICLE EXTERIORS
- F21S9/00—Lighting devices with a built-in power supply; Systems employing lighting devices with a built-in power supply
- F21S9/02—Lighting devices with a built-in power supply; Systems employing lighting devices with a built-in power supply the power supply being a battery or accumulator
- F21S9/03—Lighting devices with a built-in power supply; Systems employing lighting devices with a built-in power supply the power supply being a battery or accumulator rechargeable by exposure to light
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F21—LIGHTING
- F21W—INDEXING SCHEME ASSOCIATED WITH SUBCLASSES F21K, F21L, F21S and F21V, RELATING TO USES OR APPLICATIONS OF LIGHTING DEVICES OR SYSTEMS
- F21W2131/00—Use or application of lighting devices or systems not provided for in codes F21W2102/00-F21W2121/00
- F21W2131/10—Outdoor lighting
- F21W2131/103—Outdoor lighting of streets or roads
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B20/00—Energy efficient lighting technologies, e.g. halogen lamps or gas discharge lamps
- Y02B20/72—Energy efficient lighting technologies, e.g. halogen lamps or gas discharge lamps in street lighting
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Automation & Control Theory (AREA)
- Power Engineering (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The present disclosure provides a method and apparatus for predicting
and controlling a photovoltaic power generation capacity by improving a
similar day. The method includes: acquiring weather forecast information of
multi-consecutive days to be predicted after a present date; obtaining,
according to historical weather information with an SVM model, a historical
power generation capacity of a photovoltaic system, and the weather forecast
information of the multi-consecutive days to be predicted, a predicted power
generation capacity of each day to be predicted; obtaining a total predicted
power generation capacity according to multiple predicted power generation
capacities; and determining a daily power consumption capacity of the
photovoltaic system according to the total predicted power generation
capacity and a power storage capacity of the photovoltaic system. The present
disclosure predicts the power generation capacity of the photovoltaic system,
and controls the power consumption capacity of the photovoltaic system
according to the predicted power generation capacity and the power storage
capacity of the photovoltaic system, thereby ensuring that the photovoltaic
system with the insufficient charging capacity can maintain the operation.
ABSTRACT DRAWING: FIG. 1
-1/4
101
Acquire weather forecast information of multi-consecutive days
to be predicted after a present date
Obtain, according to historical weather information with an
SVM model, a historical power generation capacity of a 102
photovoltaic system, and the weather forecast information of the
multi-consecutive days to be predicted, a predicted power
generation capacity of each day to be predicted
103
Obtain a total predicted power generation capacity according to
multiple predicted power generation capacities
Determine a daily power consumption capacity of the 104
photovoltaic system according to the total predicted power
generation capacity and a power storage capacity of the
photovoltaic system
FIG. 1
Description
-1/4
101 Acquire weather forecast information of multi-consecutive days to be predicted after a present date
Obtain, according to historical weather information with an SVM model, a historical power generation capacity of a 102 photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted
103 Obtain a total predicted power generation capacity according to multiple predicted power generation capacities
Determine a daily power consumption capacity of the 104 photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system
FIG. 1
[01] The present disclosure relates to the technical field of solar energy, and in particular, to a method and apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day.
[02] Photovoltaic systems (such as solar streetlamps) have the advantages of environmental protection, energy conservation, no need for pipe pavement and manual operation, etc. However, due to uncertainty in weather, there is also an uncertain charging capacity of accumulators every day in the photovoltaic systems. In case of continuous overcast and rainy days, the insufficient charging capacity of the accumulators and the normal power consumption of loads of the photovoltaic systems make stored electric energy used up quickly to cause an operation failure of the photovoltaic systems, thereby greatly affecting our life and production. For example, the solar streetlamps cannot provide lighting to increase the potential safety hazards. Therefore, there is a need to monitor the photovoltaic systems remotely with communication means, such that the photovoltaic systems with the insufficient charging capacity can still maintain the operation.
[03] An objective of the present disclosure is to provide a method and apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day, which can predict a power generation capacity of the photovoltaic system, and control a power consumption capacity of the photovoltaic system according to the predicted power generation capacity and a power storage capacity of the photovoltaic system, thereby ensuring that the photovoltaic system with the insufficient charging capacity can maintain the operation.
[04] To achieve the above objective, the present disclosure provides the following solutions:
[05] A method for predicting and controlling a photovoltaic power generation capacity by improving a similar day includes:
[06] acquiring weather forecast information of multi-consecutive days to be predicted after a present date;
[07] obtaining, according to historical weather information with a support vector machine (SVM) model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted;
[08] obtaining a total predicted power generation capacity according to multiple predicted power generation capacities; and
[09] determining a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
[10] An apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day includes:
[11] a weather forecast information acquisition module, configured to acquire weather forecast information of multi-consecutive days to be predicted after a present date;
[12] a predicted power generation capacity computation module, configured to obtain, according to historical weather information with an SVM model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted;
[13] a total predicted power generation capacity determination module, configured to obtain a total predicted power generation capacity according to multiple predicted power generation capacities; and
[14] a daily power consumption capacity determination module, configured to determine a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
[15] According to specific embodiments provided in the present disclosure, the present disclosure discloses the following technical effects:
[16] The present disclosure provides a method and apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day. The method includes: acquiring weather forecast information of multi-consecutive days to be predicted after a present date; obtaining, according to historical weather information with an SVM model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted; obtaining a total predicted power generation capacity according to multiple predicted power generation capacities; and determining a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system. The present disclosure predicts the power generation capacity of the photovoltaic system, and controls the power consumption capacity of the photovoltaic system according to the predicted power generation capacity and the power storage capacity of the photovoltaic system, thereby ensuring that the photovoltaic system with the insufficient charging capacity can maintain the operation.
[17] In order to explain the technical solutions in embodiments of the present disclosure or in the prior art more clearly, the accompanying drawings required in the embodiments will be described below briefly. Apparently, the accompanying drawings in the following description show merely some embodiments of the present disclosure, and other drawings can be derived from these accompanying drawings by a person of ordinary skill in the art without creative efforts.
[18] FIG. 1 is a flow chart of a method for predicting and controlling a photovoltaic power generation capacity by improving a similar day provided by an embodiment of the present disclosure.
[19] FIG. 2 is a schematic view of a method for controlling a solar streetlamp provided by an embodiment of the present disclosure.
[20] FIG. 3 is a comparison diagram in a prediction result between a model for predicting and controlling a photovoltaic power generation capacity by improving a similar day and a comparative model provided by an embodiment of the present disclosure.
[21] FIG. 4 is a structural schematic view of an apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day provided by an embodiment of the present disclosure.
[22] The technical solutions in embodiments of the present disclosure will be described below clearly and completely with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are merely a part rather than all of the embodiments of the present disclosure. All other embodiments obtained by the person of ordinary skill in the art based on the embodiments of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.
[23] An objective of the present disclosure is to provide a method and apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day, which can predict a power generation capacity of the photovoltaic system, and control a power consumption capacity of the photovoltaic system according to the predicted power generation capacity and a power storage capacity of the photovoltaic system, thereby ensuring that the photovoltaic system with the insufficient charging capacity can maintain the operation.
[24] To make the foregoing objective, features, and advantages of the present disclosure clearer and more comprehensible, the present disclosure will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[25] FIG. 1 is a flow chart of a method for predicting and controlling a photovoltaic power generation capacity by improving a similar day provided by an embodiment of the present disclosure. As shown in FIG. 1, the present disclosure provides a method for predicting and controlling a photovoltaic power generation capacity by improving a similar day, including the following steps:
[26] Step 101: Acquire weather forecast information of multi-consecutive days to be predicted after a present date.
[27] Step 102: Obtain, according to historical weather information with an SVM model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted.
[281 Step 102 specifically includes:
[29] Determine, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted, specifically including: Construct a predicted meteorological feature vector of the day to be predicted according to the weather forecast information of the day to be predicted; construct a historical meteorological feature vector of each historical day according to historical weather information of each historical day within a preset historical time period; respectively compute a similarity between each historical day and the day to be predicted according to the predicted meteorological feature vector and multiple historical meteorological feature vectors by using a formula K
fl 1s (k) k=1 ;and sort multiple similarities in a descending manner, and determine historical days corresponding to a preset number of the similarities as the multiple historical days having the weather type same as the day to be F. predicted, where, F is a similarity between a jth historical day and the day to be predicted, K is the total number of dimensions in the predicted
meteorological feature vector, and ej(k) is a correlation coefficient between a jth historical meteorological feature vector and a kth dimension in the predicted meteorological feature vector, min min |xo (k) - x,(k) +pminmin|x,( k) -x(k) | e =k (k)x |j (k) - x,(k) |+pmin minmx(k)- k Ux,(k) | , x0(k) being the
kth dimension in the predicted meteorological feature vector, x(k) 1 being a kth dimension in the jth historical meteorological feature vector, and P being a resolution ratio.
[30] Train the SVM model by taking historical weather information corresponding to the multiple historical days having the weather type same as the day to be predicted as an input, and historical power generation capacities of the photovoltaic system corresponding to the multiple historical days having the weather type same as the day to be predicted as an output, to obtain a trained SVM model.
[31] Input the weather forecast information of the day to be predicted to the trained SVM model to obtain a predicted power generation capacity of the day to be predicted.
[32] Update the day to be predicted and return to the step of "determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted", until all days to be predicted are traversed to obtain the predicted power generation capacity of each day to be predicted.
[33] Step 103: Obtain a total predicted power generation capacity according to multiple predicted power generation capacities.
[34] Step 104: Determine a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
[35] With a solar streetlamp as an example, the present disclosure will be described below in detail. The solar light emitting diode (LED) streetlamp is applied widely because of the high luminous efficiency, environmental protection, energy conservation, no need for pipe pavement and manual operation, and so on. However, depending on weather conditions, there is an uncertain charging capacity of the solar streetlamp every day. In case of continuous overcast and rainy days, due to the insufficient charging capacity of an accumulator, the solar LED streetlamp will have the electric energy used up quickly and thus cannot provide normal lighting. In this regard, communication means are used to monitor the solar LED streetlamp remotely, as shown in FIG. 2. The remote data monitoring center establishes, through historical weather information and historical power generation capacity information of a solar panel, a power generation capacity prediction model of the solar panel, predicts a total power generation capacity of the solar LED streetlamp within future 5 days in combination with weather forecast information from a meteorological station, and sends the predicted power generation capacity to the solar LED streetlamp controller through a data communication network. The solar LED streetlamp controller computes a power consumption capacity per day according to the total predicted power generation capacity and in combination with a residual capacity of an accumulator, and adjusts a brightness, so as to guarantee lighting time with the control on the brightness of the streetlamp, thereby implementing optimal management on a lighting state of the solar LED streetlamp, and making street lighting more efficient and stable.
[36] The present disclosure determines the total power generation capacity of the solar LED streetlamp in the future 5 days by predicting the daily power generation capacity of the solar LED streetlamp. There may be two methods for predicting an output power of the photovoltaic power generation system: indirect prediction and direct prediction. The indirect prediction method is to predict a solar radiation intensity of the earth's surface first, and then obtain the output power of the system according to an output model of the photovoltaic power generation system. The method depends on solar radiation intensity information. It is not applicable to predicting the daily power generation capacity of the solar LED streetlamp because meteorological information provided by the meteorological department lacks light radiation data. The direct prediction method directly predicts the output power with historical output power data of the photovoltaic power generation system and weather forecast information. It divides weather types into a sunny day, an overcast day, a cloudy day, and a rainy day, selects a historical day similar to the day to be predicted as a training data set, and establishes a corresponding prediction model with an Elman neutral network. The daily power generation capacity of the photovoltaic system is affected by a number of meteorological factors, with the prediction accuracy being tied closely to the weather condition. In view of the demand of the off-grid photovoltaic system such as the solar LED streetlamp on accurate prediction of the daily power generation capacity, the present disclosure analyzes the meteorological factors affecting the daily power generation capacity of the photovoltaic system, and provides a method for quantizing a weather type of a day to improve the method for selecting a similar day. It predicts the daily power generation capacity of the photovoltaic system with an SVM regression method.
[37] 1. Selection on the similar day
[381 1.1 Main factors affecting the power generation capacity of the photovoltaic system
[391 The daily power generation capacity of the photovoltaic system is directly determined by the output power during power generation. The engineering model of the output power of the photovoltaic system is
P = 1 SI [1 - 0. 005 (To + 25)1 , where 7Pv is a conversion efficiency of the photovoltaic array, S is an area of the photovoltaic cell array, I is a solar
radiation intensity and tied closely to the meteorological factor, and is an environmental temperature.
[401 When the daily power generation capacity of the existing photovoltaic device is predicted, considerations may not be given to the conversion efficiency, mounting angle and the like of the photovoltaic array, because they are unchanged as constants and their influences have been implied in historical power generation capacity data. In order to extract meteorological factors greatly correlated with the daily power generation capacity of the photovoltaic system, a Pearson correlation coefficient is used to analyze the correlation between the power generation capacity system of the solar LED streetlamp and concurrent meteorological data of the local meteorological station. With the temperature T, wind speed w, humidity H, air pressure p, total cloud amount c and power generation capacity F as variables, and the timescale as 3 h, the results are as shown in Table 1. The power generation capacity of the solar LED streetlamp exhibits a strong positive correlation with the temperature, and a strong negative correlation with the humidity and the total cloud amount. Concerning the influences of the weather type and season on the generated power of the photovoltaic system, the season type is tied closely to the environmental temperature, while the weather type is tied closely to the total cloud amount. Therefore, the similar day consistent with the day to be predicted in season type and weather type is selected based on the total cloud amount, environmental temperature and environmental humidity.
[41] Table 1 Variable correlation table Variable T w H p c F T 1 0.245 0.414 0.178. 0.197 0.622 w 0.245 1 0.268 0.149 0.223 0.305 H 0.414 0.268 1 0.747 0.565 0.785 p 0.178. 0.149 0.747 1 0.457 0.349 c 0.197 0.223 0.565 0.457 1 0.554 F 0.622 0.305 0.785 0.349 0.554 [1
[42] 1.2 Selection principle of the similar day and determination of training samples
[43] 1) Construct a meteorological feature vector.
[44] The total cloud amount refers to the fraction of the sky covered by all clouds in observation. A statistical analysis made on a mass of effective historical power generation capacity data according to the database of the photovoltaic system indicates that: When the total cloud amount is up to 90%, the photovoltaic system presents the minimal power generation capacity; and the power generation capacity of the photovoltaic system in cloudy and partly cloudy days is similar to that in the sunny days. Therefore, the influence of the total cloud amount on the power generation capacity of the photovoltaic system is nonlinear. The present disclosure proposes the solution in which the weather type of a day is quantized with the total cloud amount. In view of the nonlinear influence of the total cloud amount on the power generation capacity of the photovoltaic system and in combination with the historical data analysis, the total cloud amount is remapped to a specific value between 0 and 1 to indicate the degree of the influence of the total cloud amount on the power generation capacity of the photovoltaic system, with the specific rule as
shown in Table 2, c being an encoded value of the total cloud amount.
[45] Table 2 Total cloud amount encoding table c <50% [50%, 70%) [70%, 90%) > 90% c 0 0.2 0.5 1
[461 The daily power generation capacity of the photovoltaic system in the sunny days follows a rule of rise, retention and decline. The weather type of
the day is quantized by defining a weighted average total cloud amount CP of that day: 6
C, = W c;( , = 5,8,11,14,17,20 jMI
[47] where, ci(j) and w(j) are respectively an encoded value of the total cloud amount of i at a jth moment and a corresponding weight, and
i = 5,8,11,14,17,20 represents a case where the total cloud amount at , 8, 11, 14, 17 and 20 o'clock every day is selected for research,
W'(I)i=5,8,11,14,17,20 [0.03,0.05,0.35,0.4,0.15,0.02]
[48] Respectively construct a meteorological feature vector of each day to
be predicted and historical day: ' = T,T j.. HPP , where, Y is
the meteorological feature vector, and T,, TMflHPCP arerespectively the average temperature, maximum temperature, minimum temperature, relative humidity, and weighted average total cloud amount every day.
[491 2) Upon determination of constituent vectors of each factor by computing a similarity and selecting the similar day, compute a correlation coefficient between the day to be predicted and each historical day with a grey correlation method, specifically as follows.
[50] Normalize each dimension of the meteorological feature vector with a "range method".
x(k)= y(k)-m(k)
[511 M(k)-m(k) , where, y(k), M(k) and m(k)arerespectivelya kth dimension, a minimum value of the kth dimension, and a maximum value of the kth dimension in the same meteorological feature vector.
[52] Compute a correlation coefficient between the historical meteorological feature vector and the predicted meteorological feature vector, minminix,(k) -x,(k) |+p minmiiix(k) -x (k) g,(k) = jnnfk ~() i k- |x,(k) - x (k) |+pmiiminix,(k) j kwh -xj(k) h ri e i (k) a correlation coefficient between a jth historical meteorological feature vector and a kth dimension in the predicted meteorological feature vector, x.(k) is the kth dimension in the predicted meteorological feature vector, xj(k) is a kth dimension in the jth historical meteorological feature vector, and P is a resolution ratio and is typically 0.5.
[53] Respectively compute a similarity between each historical day and the K
F1 =J7J(k) day to be predicted with a formula k=1 ; and select multiple historical days having the similarity more than a similarity threshold as similar days of the day to be predicted to form a training set, and respectively determine
similar days for 5 day to be predicteds, where, Fi is a similarity between a jth historical day and the day to be predicted, and K the total number of dimensions in the predicted meteorological feature vector, K=5. By defining the similarity with continued multiplication, the leading factors may be identified simply and automatically, and the weight setting problem of each factor may be solved.
[54] 3 SVM regression prediction model
[55] 3.1 Model training
[561 The training data set of the prediction model is the similar days of the day to be predicted and is the small samples. The SVM model is selected to establish the daily power generation capacity prediction model, because it has the small error in the solving problem with the limited training samples, overcomes the defects of long training time, poor generalization, proneness to local minimization and so on of the artificial neural network, and considers both the training error and the generalization.
[57] The support vector regression (SVR) method is used in function
regression to give the sample size i (i=1, 2, ..., m, m being the sample
size, being an input vector, and i being output data of the target function). In view of the nonlinear relationship of the sample, it is estimated that the function f(x'w) is determined with the following method: Map each sample point to a high-dimensional feature space with a nonlinear function CD(x), and then perform linear regression in the high-dimensional feature space, thereby achieving the effect of performing the linear regression in the original space.
[58] The function is:
f(xYw) = wO (x) +b = (w,'P(x)) +b
[591 where, W is a weight vector, and bi is a parameter. In order to find
w and bi , introduce slack variables i and and minimize a target
R ~ Rs = + c (;+g function. The target function R is: =2X.
[601 The constraint for the target function is:
wT0(x) + bi y: e +
s Yt - W- ~ +
[61] where, E is a coefficient of accuracy, and W is a transposition of W.
[621 At last, introduce a lagrangian multiplier, and convert the above problem into the following dual problem with a Wolf duality technique.
f (x,aa=$(ai, ai*)K(x, x)+b, fXW) may be rewritten as: i=1 , where, K(x,x,) is a kernel function, and typically a polynomial kernel function or a
radial kernel function, and aj,a are variables introduced in the support
vector algorithm. Any solution of f(x,a,,a,*) is the globally optimal solution, so there is no local extremum problem.
[631 3.2 Prediction on the daily power generation capacity
[64] According to the above analysis, the SVM model is established with the similar days of the day to be predicted as the training data set, to predict the daily power generation capacity of the photovoltaic system at the day to be predicted. The input of the SVM is the temperature every 3 h between 5:00 and 20:00 (at 5, 8, 11, 14, 17 and 20 o'clock) on the similar day and the mean daily humidity, and there are 13 input variables. The output is the correspondingly daily power generation capacity, and there is 1 output variable. Specifically:
[65] 1) Smooth historical data of the photovoltaic system and eliminate singular data therein.
[66] 2) Establish a training sample set and a prediction sample set, and normalize sample data.
[67] 3) Train selected parameters c, E and 5 of the SVM model with a radial basis function as a kernel function, and predict a photovoltaic output of the day to be predicted with the trained model.
[68] 4. Result analysis
[69] With historical data and meteorological data of the 100 W solar LED streetlamp at some place between July 1, 2016 and June 30, 2017 (1 year) as the samples, and in combination with weather information on the day to be predicted, the SVM regression prediction model is established to obtain the predicted value of the daily power generation capacity. In order to verify the influences of the remapping and encoding of the total cloud amount as well as the effectiveness and accuracy of the model for predicting the daily power generation capacity of the photovoltaic system, the prediction model provided
by the present disclosure is used as M0 , and the comparative model 1,
similar day-Elman neutral network prediction model M2, single SVR
prediction model M3 and self-organizing map (SOM)-back-propagation (BP)
neutral network prediction model M4 that are not subjected to the remapping of the total cloud amount are added. The root-mean-square error
eRMSE and the mean absolute percentage error eMAPE are used to evaluate predictive abilities of the 5 prediction models on the daily power generation capacity.
[70] Specifically, the root-mean-square error RMSE and the mean absolute
percentage erroreMAPE are computed as follows:
1 M(P - P) i eN P
[71] where, N is the number of historical days included in the test set, and
P and the are respectively the predicted value and the actual value,
[72] Specifically, the predictive abilities of 5 prediction models on the daily power generation capacity are as shown in Table 3.
[731 Table 3 Comparison table on prediction accuracy of models MO -4
Indicator MM M M2 M3 M4
eMSE 0.0350 0.0390 0.0590 0.0540 0.0610
[74] As can be seen from Table 3, the two errors of the M0 and M1 are far less than those of rest models, and the SVM model based on the similar days achieves the desirable modeling effect. The comparison diagram in prediction
result between M0 and M' is as shown in FIG. 3. In FIG. 3, the more concentrated the test data points near the diagonal line, the better the prediction effect of the model. On the basis of FIG. 3, fitting straight lines of
the models M 0 and M1 are obtained through linear regression:
[75] M m odel: Predicted value=Actual value*0.92+0. 01
[76] M1 model: Predicted value=Actual value*0.88+0. 02
[77] As can be seen, the fitting straight line of the model MO has the slope closer to 1, and the data points are more concentrated near the diagonal line. Therefore, the prediction model with the remapping and encoding of the total cloud amount has a higher precision accuracy.
[781 In view of the demand of the off-grid photovoltaic system such as the solar LED streetlamp on accurate prediction of the daily power generation capacity, the present disclosure mainly analyzes the influence of the total cloud amount on the power generation capacity. By quantizing the weather type of the day, and reconstructing the weather feature vector during selection of the similar day in combination with many meteorological factors, the present disclosure establishes the SVR prediction model to predict the daily power generation capacity of the photovoltaic system. By verifying the model with 1-year test data of the solar LED streetlamp in some place, the results indicate that the prediction model achieves the desirable accuracy, and has the feasibility and practicability.
[79] FIG. 4 is a structural schematic view of an apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day provided by an embodiment of the present disclosure. As shown in FIG. 4, the present disclosure further provides an apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day, including: a weather forecast information acquisition module 401, a predicted power generation capacity computation module 402, a total predicted power generation capacity determination module 403, and a daily power consumption capacity determination module 403.
[80] The weather forecast information acquisition module 401 is configured to acquire weather forecast information of multi-consecutive days to be predicted after a present date.
[81] The predicted power generation capacity computation module 402 is configured to obtain, according to historical weather information with an SVM model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted.
[82] The predicted power generation capacity computation module 402 specifically includes: a historical day determination unit, an SVM model training unit, a predicted power generation capacity computation unit, and an update unit.
[83] The historical day determination unit is configured to determine, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted.
[84] The historical day determination unit specifically includes:
[85] a predicted meteorological feature vector construction subunit, configured to construct a predicted meteorological feature vector of the day to be predicted according to the weather forecast information of the day to be predicted;
[86] a historical meteorological feature vector construction subunit, configured to construct a historical meteorological feature vector of each historical day according to historical weather information of each historical day within a preset historical time period;
[871 a similarity computation subunit, configured to respectively compute a similarity between each historical day and the day to be predicted according to the predicted meteorological feature vector and multiple historical K F.= 7j(k) meteorological feature vectors by using a formula k=1 ; and
[881 a sorting subunit, configured to sort multiple similarities in a descending manner, and determine historical days corresponding to a preset number of the similarities as the multiple historical days having the weather type same as the day to be predicted;
[891 where, F' is a similarity between a jth historical day and the day to be predicted, K is the total number of dimensions in the predicted meteorological
feature vector, and ej(k) is a correlation coefficient between a jth historical meteorological feature vector and a kth dimension in the predicted meteorological feature vector, minx IX(k) - x(k) +pmin minx(k) - x (k)
IxO(k) - x,(k) |+p minmi |x (k) - x,(k) | x (k) being the
kth dimension in the predicted meteorological feature vector, xj(k) being a kth dimension in the jth historical meteorological feature vector, and P being a resolution ratio.
[90] The SVM model training unit is configured to train the SVM model by taking historical weather information corresponding to the multiple historical days having the weather type same as the day to be predicted as an input, and historical power generation capacities of the photovoltaic system corresponding to the multiple historical days having the weather type same as the day to be predicted as an output, to obtain a trained SVM model.
[91] The predicted power generation capacity computation unit is configured to input the weather forecast information of the day to be predicted to the trained SVM model to obtain a predicted power generation capacity of the day to be predicted.
[92] The update unit is configured to update the day to be predicted and return to the step of "determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted", until all days to be predicted are traversed to obtain the predicted power generation capacity of each day to be predicted.
[93] The total predicted power generation capacity determination module 403 is configured to obtain a total predicted power generation capacity according to multiple predicted power generation capacities.
[94] The daily power consumption capacity determination module 403 is configured to determine a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
[95] Each embodiment of the present specification is described in a progressive manner, each embodiment focuses on the difference from other embodiments, and the same and similar parts between the embodiments may refer to each other. Since the system disclosed in the embodiments corresponds to the method disclosed in the embodiments, the description is relatively simple, and reference can be made to the method description.
[96] In this specification, several specific embodiments are used for illustration of the principles and implementations of the present disclosure. The description of the foregoing embodiments is used to help illustrate the method of the present disclosure and the core ideas thereof. In addition, those of ordinary skill in the art can make various modifications in terms of specific implementations and the scope of application in accordance with the ideas of the present disclosure. In conclusion, the content of this specification shall not be construed as a limitation to the present disclosure.
Claims (5)
1. A method for predicting and controlling a photovoltaic power generation capacity by improving a similar day, comprising: acquiring weather forecast information of multi-consecutive days to be predicted after a present date; obtaining, according to historical weather information with a support vector machine (SVM) model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted; obtaining a total predicted power generation capacity according to multiple predicted power generation capacities; and determining a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
2. The method for predicting and controlling a photovoltaic power generation capacity by improving a similar day according to claim 1, wherein the obtaining, according to historical weather information with an SVM model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted specifically comprises: determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted; training the SVM model by taking historical weather information corresponding to the multiple historical days having the weather type same as the day to be predicted as an input, and historical power generation capacities of the photovoltaic system corresponding to the multiple historical days having the weather type same as the day to be predicted as an output, to obtain a trained SVM model; inputting the weather forecast information of the day to be predicted to the trained SVM model to obtain a predicted power generation capacity of the day to be predicted; and updating the day to be predicted and returning to the step of "determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted", until all days to be predicted are traversed to obtain the predicted power generation capacity of each day to be predicted.
3. The method for predicting and controlling a photovoltaic power generation capacity by improving a similar day according to claim 2, wherein the determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted specifically comprises: constructing a predicted meteorological feature vector of the day to be predicted according to the weather forecast information of the day to be predicted; constructing a historical meteorological feature vector of each historical day according to historical weather information of each historical day within a preset historical time period; respectively computing a similarity between each historical day and the day to be predicted according to the predicted meteorological feature vector and multiple historical meteorological feature vectors by using a formula K F,=J e7j(k) k=1 ; and sorting multiple similarities in a descending manner, and determining historical days corresponding to a preset number of the similarities as the multiple historical days having the weather type same as the day to be predicted,
wherein, Fi is a similarity between a jth historical day and the day to be predicted, K is the total number of dimensions in the predicted meteorological
feature vector, and j(k) is a correlation coefficient between a jth historical meteorological feature vector and a kth dimension in the predicted meteorological feature vector, min minxo(k) - xJ(k) +pminmin x(k) -x(k)
Ix. (k) - xj(k) |+pmin i minjxO(k) - )(k k being the
kth dimension in the predicted meteorological feature vector, xj(k) being a kth dimension in the jth historical meteorological feature vector, and P being a resolution ratio.
4. An apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day, comprising: a weather forecast information acquisition module, configured to acquire weather forecast information of multi-consecutive days to be predicted after a present date; a predicted power generation capacity computation module, configured to obtain, according to historical weather information with a support vector machine (SVM) model, a historical power generation capacity of a photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted; a total predicted power generation capacity determination module, configured to obtain a total predicted power generation capacity according to multiple predicted power generation capacities; and a daily power consumption capacity determination module, configured to determine a daily power consumption capacity of the photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system.
5. The apparatus for predicting and controlling a photovoltaic power generation capacity by improving a similar day according to claim 4, wherein the predicted power generation capacity computation module specifically comprises: a historical day determination unit, configured to determine, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted; an SVM model training unit, configured to train the SVM model by taking historical weather information corresponding to the multiple historical days having the weather type same as the day to be predicted as an input, and historical power generation capacities of the photovoltaic system corresponding to the multiple historical days having the weather type same as the day to be predicted as an output, to obtain a trained SVM model; a predicted power generation capacity computation unit, configured to input the weather forecast information of the day to be predicted to the trained SVM model to obtain a predicted power generation capacity of the day to be predicted; and an update unit, configured to update the day to be predicted and return to the step of "determining, according to weather forecast information of a day to be predicted, multiple historical days having a weather type same as the day to be predicted", until all days to be predicted are traversed to obtain the predicted power generation capacity of each day to be predicted.
－1/4－ 22 Jul 2021
101 Acquire weather forecast information of multi-consecutive days to be predicted after a present date
Obtain, according to historical weather information with an SVM model, a historical power generation capacity of a 102 2021104436
photovoltaic system, and the weather forecast information of the multi-consecutive days to be predicted, a predicted power generation capacity of each day to be predicted
103 Obtain a total predicted power generation capacity according to multiple predicted power generation capacities
Determine a daily power consumption capacity of the 104 photovoltaic system according to the total predicted power generation capacity and a power storage capacity of the photovoltaic system
FIG. 1
－2/4－ 22 Jul 2021
Predicted power Weather Remote data generation information monitoring capacity in in future 5 and power future 5 days Solar LED Meteorological days generation streetlamp control station capacity terminal prediction Working 2021104436
center (PC state data terminal) of streetlamp
FIG. 2
－3/4－ 22 Jul 2021
Predicted value of daily power generation capacity / kW·h 2021104436
M0 model M1 model
Actual value of daily power generation capacity / kW·h
FIG. 3
－4/4－ 22 Jul 2021
401 Weather forecast information acquisition module
Predicted power generation capacity computation 402 2021104436
module
403 Total predicted power generation capacity determination module
404 Daily power consumption capacity determination module
FIG. 4
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021104436A AU2021104436A4 (en) | 2021-07-22 | 2021-07-22 | Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021104436A AU2021104436A4 (en) | 2021-07-22 | 2021-07-22 | Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| AU2021104436A4 true AU2021104436A4 (en) | 2021-09-16 |
Family
ID=77666583
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| AU2021104436A Active AU2021104436A4 (en) | 2021-07-22 | 2021-07-22 | Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day |
Country Status (1)
| Country | Link |
|---|---|
| AU (1) | AU2021104436A4 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114362175A (en) * | 2022-03-10 | 2022-04-15 | 山东大学 | Wind power prediction method and system based on depth certainty strategy gradient algorithm |
| WO2023222401A1 (en) | 2022-05-19 | 2023-11-23 | IFP Energies Nouvelles | Method for forecasting a power produced by at least one photovoltaic panel |
-
2021
- 2021-07-22 AU AU2021104436A patent/AU2021104436A4/en active Active
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114362175A (en) * | 2022-03-10 | 2022-04-15 | 山东大学 | Wind power prediction method and system based on depth certainty strategy gradient algorithm |
| CN114362175B (en) * | 2022-03-10 | 2022-06-07 | 山东大学 | Wind power prediction method and system based on depth certainty strategy gradient algorithm |
| WO2023222401A1 (en) | 2022-05-19 | 2023-11-23 | IFP Energies Nouvelles | Method for forecasting a power produced by at least one photovoltaic panel |
| FR3135798A1 (en) | 2022-05-19 | 2023-11-24 | IFP Energies Nouvelles | Method for predicting power produced by at least one photovoltaic panel |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2021104436A4 (en) | Method and apparatus for predicting and controlling photovoltaic power generation capacity by improving similar day | |
| CN109884896B (en) | Photovoltaic tracking system optimized tracking method based on similar day irradiation prediction | |
| CN112785027B (en) | Wind-solar-storage combined power generation system confidence capacity evaluation method and system | |
| KR101597993B1 (en) | Energy storage system | |
| CN110929953A (en) | Photovoltaic power station ultra-short term output prediction method based on cluster analysis | |
| CN103699944A (en) | Wind and photovoltaic generation power prediction system with multiple prediction modes | |
| Tukymbekov et al. | Intelligent autonomous street lighting system based on weather forecast using LSTM | |
| KR102230548B1 (en) | Power generation prediction and efficiency diagnosis system of solar power generation facilities using FRBFNN model | |
| CN111191854A (en) | Photovoltaic power generation prediction model and method based on linear regression and neural network | |
| Park et al. | Multi-layer RNN-based short-term photovoltaic power forecasting using IoT dataset | |
| KR20200119367A (en) | Demand power prediction device for energy storage system and method for predicting demand power using the same | |
| Eniola et al. | Hour-ahead forecasting of photovoltaic power output based on hidden Markov model and genetic algorithm | |
| Arévalo et al. | Smart monitoring method for photovoltaic systems and failure control based on power smoothing techniques | |
| Meng et al. | A new PV generation power prediction model based on GA-BP neural network with artificial classification of history day | |
| Yan et al. | Operating power reserve quantification through PV generation uncertainty analysis of a microgrid | |
| Ku et al. | Building electric energy prediction modeling for BEMS using easily obtainable weather factors with Kriging model and data mining | |
| CN110991747A (en) | Short-term load prediction method considering wind power plant power | |
| CN117787744A (en) | Power generation amount abnormality detection system and method based on prediction algorithm | |
| KR102299698B1 (en) | Power Consumption Prediction System Using Long-Term and Short-Term Prediction method | |
| KR102153722B1 (en) | PV Power Generation Management Apparatus and Method | |
| Matushkin et al. | Analysis of factors for forecasting electric power generation by solar power plants | |
| Miao et al. | Harnessing climate variables for predicting PV power output: A backpropagation neural network analysis in a subtropical climate region | |
| CN116484998A (en) | Distributed photovoltaic power station power prediction method and system based on meteorological similar day | |
| CN112116127B (en) | Photovoltaic power prediction method based on association of meteorological process and power fluctuation | |
| Belloni et al. | A street lighting based on solar energy: optimization of a neural network for the productivity evaluation in different climate conditions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FGI | Letters patent sealed or granted (innovation patent) |