CN117350157A - Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method - Google Patents
Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method Download PDFInfo
- Publication number
- CN117350157A CN117350157A CN202311319910.1A CN202311319910A CN117350157A CN 117350157 A CN117350157 A CN 117350157A CN 202311319910 A CN202311319910 A CN 202311319910A CN 117350157 A CN117350157 A CN 117350157A
- Authority
- CN
- China
- Prior art keywords
- irradiance
- time
- power
- photovoltaic power
- power station
- 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.)
- Pending
Links
- 238000013507 mapping Methods 0.000 title claims abstract description 69
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 9
- 230000002123 temporal effect Effects 0.000 claims description 9
- 238000010586 diagram Methods 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 230000001364 causal effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000012098 association analyses Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000005855 radiation Effects 0.000 claims description 3
- 238000005314 correlation function Methods 0.000 abstract 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Artificial Intelligence (AREA)
- Water Supply & Treatment (AREA)
- General Business, Economics & Management (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- Power Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a space-time mapping region irradiance estimation and distributed photovoltaic power prediction method, which comprises the steps of obtaining irradiance of a selected power station and distributed photovoltaic power station power data; constructing an irradiation intensity and distributed photovoltaic power station power output correlation function model, and obtaining irradiance of a selected power station; constructing an extraterrestrial irradiance and surface irradiance mapping relation model, dividing weather modes into a sunny day and a cloudy day, and respectively constructing the sunny day model and the cloudy day model to obtain irradiance estimation values under different modes; calculating a cross-correlation coefficient of irradiance sequences, and selecting an associated power station; calculating irradiance estimation values of the positions of the selected associated power stations by adopting an extraterrestrial irradiance and surface irradiance mapping relation model; training a built irradiation intensity and distributed photovoltaic power station power output correlation function model, and calculating a power estimated value; constructing a space-time double-scale network, and predicting photovoltaic power in a future set time; the prediction accuracy of the invention is improved, and the practical applicability is enhanced.
Description
Technical Field
The invention belongs to the technical field of electric automation, and particularly relates to a space-time mapping region irradiance estimation and distributed photovoltaic power prediction method.
Background
At present, a large-scale distributed photovoltaic is connected into a power grid, so that the relative shortage of power in partial remote areas is relieved, and meanwhile, a plurality of adverse effects are brought to the safe and stable operation of a power system.
The net load prediction plays an important role in maintaining the stability of the power system, and a large number of researches show that when the power consumption and the power output in the power grid are decomposed and respectively predicted, the prediction precision can be improved; however, at present, no special power prediction system is required to be configured for the distributed photovoltaic, and related weather or cloud image observation facilities are not specially equipped for the reasons of small single capacity, scattered installation, small investment and the like, so that the distributed photovoltaic power prediction generally only depends on the power data of the distributed photovoltaic, and the condition of directly applying the centralized photovoltaic power prediction method is not provided.
In summary, the current irradiance estimation distributed photovoltaic power prediction based on the space-time mapping region has the problems of low precision and weak practical applicability.
Disclosure of Invention
The invention aims to provide a space-time mapping region irradiance estimation and distributed photovoltaic power prediction method with improved prediction precision and enhanced practical applicability.
The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method provided by the invention comprises the following steps:
s1, irradiance of a selected power station and power data of a distributed photovoltaic power station are obtained;
s2, constructing an irradiance and distributed photovoltaic power station power output mapping model by adopting the data obtained in the step S1, and obtaining irradiance of a selected power station;
s3, constructing an extraterrestrial irradiance and surface irradiance mapping relation model of the selected area, dividing weather modes into a sunny day and a cloudy day at the same time, respectively constructing the sunny day mapping relation model and the cloudy day mapping relation model according to the divided weather modes, and acquiring irradiance estimation values of set positions in different modes through the mapping relation model;
s4, calculating a cross-correlation coefficient of the irradiance sequence, and selecting an associated power station which meets the set condition in the set area;
s5, calculating irradiance estimation values of the positions of the selected associated power stations by adopting the map relation model of the external irradiance and the surface irradiance constructed in the step S3;
s6, training the irradiance and distributed photovoltaic power station power output mapping model constructed in the step S2 by adopting the data acquired in the step S1, and calculating a power estimated value;
s7, constructing a space-time double-scale network by adopting the power station selected in the step S4, the irradiance estimation value calculated in the step S5 and the power estimation value calculated in the step S6, and predicting the photovoltaic power in the future set time;
the step S2 of constructing an irradiance and distributed photovoltaic power station power output mapping model by adopting the data acquired in the step S1, and acquiring irradiance of a selected power station specifically comprises the following steps:
the irradiation intensity is related to the power output of the distributed photovoltaic power station by adopting the following formula:
wherein,an estimate representing the power produced by a given photovoltaic power plant; t represents the number of time steps in the data; />Is indicated at +.>A vector of irradiance observed thereat; alpha represents a vector of the inclination angle of the photovoltaic module; beta represents a vector containing the azimuth of the photovoltaic module; g represents a vector containing the geographical coordinates of the power station, including latitude, longitude, altitude; />Is indicated at +.>A vector of temperatures observed at; p (P) nom A vector representing the nominal power of the photovoltaic module;
different irradiance value functions can return the same power output, and the following steps are adopted for processing:
(1) Estimating a panel orientation from the measured power of a given photovoltaic power plant;
(2) Constructing a function f by using the calculated direction, and solving;
the following formula is used to calculate the optimum value of irradiance:
wherein IRRopt represents an optimized value of irradiance; p represents a photovoltaic power observation;
solving the above formula by using one or several photovoltaic power values is as follows:
wherein n is pv Representing the total number of photovoltaic power signals;representing the observed value after normalization processing; />Representing the normalized power estimation value;
the step S3 is to construct a mapping relation model of the irradiance outside the ground and the irradiance on the surface of the selected area, divide the weather modes into a sunny day and a cloudy day at the same time, respectively construct a mapping relation model of the sunny day and a mapping relation model of the cloudy day according to the divided weather modes, and obtain irradiance estimation values of set positions under different modes through the mapping relation model, and specifically comprises the following steps:
dividing the inverted irradiance value in the set time into a sunny day or a cloudy day; the dividing method adopts five standards: an average value; a maximum value; the line length of irradiance versus time curve; standard deviation of irradiance change rate; the maximum difference between irradiance variation and clear sky time sequence;
constructing a map model of the extraterrestrial irradiance and the surface irradiance of the selected area; the method comprises the steps of selecting external irradiance, ambient temperature, solar altitude angle and solar azimuth angle as input, taking surface irradiance as output, and respectively constructing a sunny day mapping relation model and a cloudy day mapping relation model according to divided weather patterns through a BP neural network;
the outside radiation illuminance at the actual plant position was calculated using the following formula:
G＝kG 0 sinh＝kG 0 cosθ z ＝kG 0 (sinδsinφ+cosδcosφcosω)
wherein k represents a solar-earth distance correction coefficient; g 0 Representing the solar constant; θ z Representing zenith angle; delta represents declination of the sun; phi represents the local latitude; omega represents the solar time angle;
the calculated irradiance maps the estimated irradiance of the actual station, and the obtained irradiance and surface irradiance mapping relation model is trained and divided into a sunny day mapping relation model and a cloudy day mapping relation model;
the step S4 of calculating the cross-correlation coefficient of irradiance sequences, selecting the associated power stations meeting the set conditions in the set area, specifically includes:
selecting an associated power station meeting a set condition through irradiance space-time association analysis:
the setting condition comprises selecting a plurality of power stations with maximum correlation coefficients according to the set number;
generating an icosahedron triangular grid on a unit sphere, and analyzing irradiance sequence characteristics and space-time association relations among power stations under different weather conditions aiming at simulated photovoltaic power stations; obtaining an associated power station meeting a set condition by calculating pearson correlation coefficients between irradiance sequences, wherein the calculation comprises sequentially recording irradiance values according to a time sequence in the set time according to irradiance values in the set time, and defining the recorded values as irradiance time sequence data; analyzing and obtaining the correlation degree of irradiance at different places in the region and the delay condition between irradiance sequences through the cross-correlation coefficient;
the calculation of the irradiance estimation value of the given associated power station position by adopting the mapping relation model of the extraterrestrial irradiance and the surface irradiance constructed in the step S3 in the step S5 specifically comprises the following steps:
mapping the ground irradiance estimation value of a given position by adopting the constructed ground irradiance and ground irradiance mapping model and the calculated ground irradiance of the given associated power station position and the trained neural network model;
the training step S2 of the irradiance and distributed photovoltaic power station power output mapping model constructed by the data acquired in the step S1, and calculating the power estimation value specifically includes:
selecting irradiance on the earth surface, ambient temperature, solar altitude angle, solar azimuth angle and historical power as input, actual power as output, and selecting BP neural network to build a model;
the power station selected in the step S4, the irradiance estimation value calculated in the step S5, and the power estimation value calculated in the step S6 in the step S7 are adopted to construct a space-time double-scale network, and the prediction is performed for the photovoltaic power in the future set time, which specifically comprises:
(7-1) constructing an undirected graph:
and (3) constructing an undirected graph consisting of a plurality of nodes by adopting the distributed photovoltaic power station selected in the step (S4), and adopting the following formula to express:
G t ＝(P t ,ε,W)
wherein G is t A graph showing time t; p (P) t The representation node set is the power set of each distributed power station at the moment t; epsilon represents the collection of edges between nodes; w represents the weight of the adjacency matrix;
(7-2) graph convolution extracts spatial features of data:
the spatial characteristics of the graph convolution extracted data are obtained through Chebyshev polynomial approximation, and the Chebyshev graph convolution formula is as follows:
wherein Θ represents any parameter and is adjusted by initializing assignment and utilizing error back propagation; g represents a graph structure; x represents an input feature; Θ (L) represents a convolution kernel; k meterDiagram convolution kernel size; θ k Representing polynomial coefficients;a polynomial expansion approximation representing Laplacian;
(7-3) time characteristics of the gated convolution extracted data:
gating convolution is used to capture temporal features in the data; setting the convolution width to K t For extracting K t-1 Time characteristics of (2);
for each node in graph G, the time convolution continually searches K of the input element in a filling manner t Neighborhood, so that the time sequence length is shortened by K each time t-1 ；
Length M, with C, to be input to each node i The time domain convolution sequence of each channel is equivalent toY maps to a single output +.>
The above process is represented by the following formula:
wherein Γ represents a convolution kernel; p represents the first half of the causal convolution into which the input is split; the product of Hadamard; sigma (·) represents that the input P controlling the current state is related to the composition structure and dynamic variance in the time series; q represents the second half of the output split representing the causal convolution;
(7-4) time series of spatio-temporal convolution block joint processing graph structure:
adopting a space-time convolution block to jointly process the time sequence of the graph structure, and fusing the characteristics of space and time;
setting the input and output of the space-time convolution block as a three-dimensional tensor;
the following general formula is adoptedCalculating the input p of the first block l Output p l+1 ：
Wherein,a lower temporal kernel representing a first block; reLU (·) represents a true streamline unit function; theta (theta) l A common kernel representing the graph convolution; />An upper temporal kernel representing a first block;
the loss function of the space-time diagram convolution is calculated using the following formula:
wherein,representing the predicted value; w (W) θ Training parameters representing the model; p is p t+1 Representing the true value; p is p j Representing the output map of the spatio-temporal block, j=t+1, …, t-m+1;
(7-5) predicting photovoltaic power for a future set time:
training the built space-time diagram convolutional neural network according to the obtained irradiance and power data of the given associated power station, taking each distributed power station as a node, calculating the distance between the nodes by using longitude and latitude, and predicting the photovoltaic power in the future set time;
according to the space-time mapping region irradiance estimation and distributed photovoltaic power prediction method provided by the invention, irradiance is estimated by using the generated power measurement of a single distributed photovoltaic power station through an unsupervised method, and a mapping model is built based on the external irradiance to obtain regional-level space-time resolution irradiance; the inverted irradiance is not only used for irradiance/photovoltaic power prediction, but also is convenient for separating photovoltaic power generation from power load, so that the accuracy of the summarized net load prediction in a low-voltage power grid is improved; the power data of a given photovoltaic power station are mapped, a space-time double-scale network is established based on a deep learning method, and space-time correlation among distributed power station groups is deeply excavated to improve the prediction precision of the generated power; the prediction accuracy of the method is improved, and the practical applicability is enhanced.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
FIG. 2 is a graph of cross-correlation coefficients between irradiance sequences of a selected photovoltaic power plant in the method of the present invention.
Detailed Description
A schematic process flow diagram of the method of the present invention is shown in fig. 1: the space-time mapping region irradiance estimation and distributed photovoltaic power prediction method provided by the invention comprises the following steps:
s1, irradiance of a selected power station and power data of a distributed photovoltaic power station are obtained;
s2, constructing a function model related to irradiation intensity and power output of the distributed photovoltaic power station by adopting the data obtained in the step S1, and obtaining irradiance of the selected power station; the method specifically comprises the following steps:
the irradiation intensity is related to the power output of the distributed photovoltaic power station by adopting the following formula:
wherein,an estimate representing the power produced by a given photovoltaic power plant; t represents the number of time steps in the data; />Is indicated at +.>A vector of irradiance observed thereat; α represents a vector containing the tilt of the module; beta represents a vector containing the azimuth of the photovoltaic module; g represents a vector containing the geographical coordinates of the power station, including latitude, longitude, altitude; />Is indicated at +.>A vector of temperatures observed at; p (P) nom A vector representing the nominal power of the photovoltaic module;
different irradiance value functions can return the same power output, and the following steps are adopted for processing:
(1) Estimating a panel orientation from the measured power of a given photovoltaic power plant;
(2) Constructing a function f by using the calculated direction, and solving;
the following formula is used to calculate the optimum value of irradiance:
wherein IRRopt represents an optimized value of irradiance; p represents a photovoltaic power observation;
solving the above formula by using one or several photovoltaic power values is as follows:
wherein n is pv Representing the total number of photovoltaic power signals;representing the observed value after normalization processing; />Representing the normalized estimated power value;
s3, constructing an extraterrestrial irradiance and surface irradiance mapping relation model of the selected area, dividing weather modes into a sunny day and a cloudy day at the same time, respectively constructing the sunny day mapping relation model and the cloudy day mapping relation model according to the divided weather modes, and acquiring irradiance estimation values of set positions in different modes through the mapping relation model; the method specifically comprises the following steps:
constructing a map model of the extraterrestrial irradiance and the surface irradiance of the selected area; the method comprises the steps of selecting external irradiance, ambient temperature, solar altitude angle and solar azimuth angle as input, taking surface irradiance as output, and respectively constructing a sunny day mapping relation model and a cloudy day mapping relation model according to divided weather patterns through a BP neural network;
the outside radiation illuminance at the actual plant position was calculated using the following formula:
G＝kG 0 sinh＝kG 0 cosθ z ＝kG 0 (sinδsinφ+cosδcosφcosω)
wherein k represents a solar-earth distance correction coefficient; g 0 Representing the solar constant, a fixed value of 1386W/m was chosen in the method of the invention 2 ；θ z Representing zenith angle, θ was selected in the method of the invention z =90 °/-h; delta represents declination of the sun; phi represents the local latitude; omega represents the solar time angle;
the calculated irradiance maps the estimated irradiance of the actual station, and the obtained irradiance and surface irradiance mapping relation model is trained and divided into a sunny day mapping relation model and a cloudy day mapping relation model;
s4, calculating a cross-correlation coefficient of the irradiance sequence, and selecting an associated power station which meets the set condition in the set area; the method specifically comprises the following steps:
selecting an associated power station meeting a set condition through irradiance space-time association analysis:
the setting condition comprises selecting a plurality of power stations with maximum correlation coefficients according to the set number; in the method, the number of the selected power stations is 10;
generating an icosahedron triangular grid on a unit sphere, and analyzing irradiance sequence characteristics and space-time association relations among power stations under different weather conditions aiming at simulated photovoltaic power stations; obtaining an associated power station meeting a set condition by calculating pearson correlation coefficients between irradiance sequences, wherein the calculation comprises sequentially recording irradiance values according to a time sequence in the set time according to irradiance values in the set time, and defining the recorded values as irradiance time sequence data; analyzing and obtaining the correlation degree of irradiance at different places in the region and the delay condition between irradiance sequences through the cross-correlation coefficient;
the cross-correlation coefficient between irradiance sequences of the photovoltaic power station selected in the method of the present invention is shown in fig. 2: in the method, three photovoltaic power stations are selected to analyze irradiance sequence characteristics and space-time association relations among all the power stations under different weather conditions in cloudy weather, and the power stations are set as A, B, C respectively; as can be seen from fig. 2, the sequence correlation coefficient of the data of the a-site and the B-site is close to 1, which indicates that the similarity between the sequences is large, and the mutual expression capability between the sequences is strong; the cross correlation coefficient of the irradiance of the A site and the irradiance of the C site is delayed for about 15min to generate the maximum value, which indicates that the correlation of the two time sequences is highest at the moment;
s5, calculating irradiance estimation values of the positions of the selected associated power stations by adopting the map relation model of the external irradiance and the surface irradiance constructed in the step S3; the method specifically comprises the following steps:
mapping the ground irradiance estimation value of a given position by adopting the constructed ground irradiance and ground irradiance mapping model and the calculated ground irradiance of the given associated power station position and the trained neural network model;
s6, training the irradiance and distributed photovoltaic power station power output mapping model constructed in the step S2 by adopting the data acquired in the step S1, and calculating a power estimated value; the method specifically comprises the following steps:
selecting irradiance on the earth surface, ambient temperature, solar altitude angle, solar azimuth angle and historical power as input, actual power as output, and selecting BP neural network to build a model;
s7, constructing a space-time double-scale network by adopting the power station selected in the step S4, the irradiance estimation value calculated in the step S5 and the power estimation value calculated in the step S6, and predicting the photovoltaic power in the future set time; the method specifically comprises the following steps:
(7-1) constructing an undirected graph:
and (3) constructing an undirected graph consisting of a plurality of nodes by adopting the distributed photovoltaic power station selected in the step (S4), and adopting the following formula to express:
G t ＝(P t ,ε,W)
wherein G is t A graph showing time t; p (P) t The representation node set is the power set of each distributed power station at the moment t; epsilon represents the collection of edges between nodes; w represents the weight of the adjacency matrix;
(7-2) graph convolution extracts spatial features of data:
the spatial characteristics of the graph convolution extracted data are obtained through Chebyshev polynomial approximation, and the Chebyshev graph convolution formula is as follows:
wherein Θ represents any parameter and is adjusted by initializing assignment and utilizing error back propagation; g represents a graph structure; x represents an input feature; Θ (L) represents a convolution kernel; k represents the size of a graph convolution kernel; θ k Representing polynomial coefficients;a polynomial expansion approximation representing Laplacian;
(7-3) time characteristics of the gated convolution extracted data:
gating convolution is used to capture temporal features in the data; setting the convolution width to K t For extracting K t-1 Time characteristics of (2);
for each node in graph G, the time convolution continually searches K of the input element in a filling manner t Neighborhood, such that time series lengthEach time shorten K t-1 ；
Length M, with C, to be input to each node i The time domain convolution sequence of each channel is equivalent toY maps to a single output +.>
The above process is represented by the following formula:
wherein Γ represents a convolution kernel; p represents the first half of the causal convolution into which the input is split; the product of Hadamard; sigma (·) represents that the input P controlling the current state is related to the composition structure and dynamic variance in the time series; q represents the second half of the output split representing the causal convolution;
(7-4) time series of spatio-temporal convolution block joint processing graph structure:
adopting a space-time convolution block to jointly process the time sequence of the graph structure, and fusing the characteristics of space and time;
setting the input and output of the space-time convolution block as a three-dimensional tensor;
calculating the input p of the first block using the following formula l Output p l+1 ：
Wherein,a lower temporal kernel representing a first block; reLU (·) represents a true streamline unit function; theta (theta) l A common kernel representing the graph convolution; />An upper temporal kernel representing a first block;
the loss function of the space-time diagram convolution is calculated using the following formula:
wherein,representing the predicted value; w (W) θ Training parameters representing the model; p is p t+1 Representing the true value; p is p j Representing the output map of the spatio-temporal block, j=t+1, …, t-m+1;
(7-5) predicting photovoltaic power for a future set time:
and training the built space-time diagram convolutional neural network according to the obtained irradiance and power data of the given associated power station, taking each distributed power station as a node, calculating the distance between the nodes by using longitude and latitude, and predicting the photovoltaic power in the future set time.
Claims (7)
1. A space-time mapping region irradiance estimation and distributed photovoltaic power prediction method comprises the following steps:
s1, irradiance of a selected power station and power data of a distributed photovoltaic power station are obtained;
s2, constructing an irradiance and distributed photovoltaic power station power output mapping model by adopting the data obtained in the step S1, and obtaining irradiance of a selected power station;
s3, constructing an extraterrestrial irradiance and surface irradiance mapping relation model of the selected area, dividing weather modes into a sunny day and a cloudy day at the same time, respectively constructing the sunny day mapping relation model and the cloudy day mapping relation model according to the divided weather modes, and acquiring irradiance estimation values of set positions in different modes through the mapping relation model;
s4, calculating a cross-correlation coefficient of the irradiance sequence, and selecting an associated power station which meets the set condition in the set area;
s5, calculating irradiance estimation values of the positions of the selected associated power stations by adopting the map relation model of the external irradiance and the surface irradiance constructed in the step S3;
s6, training the irradiance and distributed photovoltaic power station power output mapping model constructed in the step S2 by adopting the data acquired in the step S1, and calculating a power estimated value;
s7, constructing a space-time double-scale network by adopting the power station selected in the step S4, the irradiance estimated value calculated in the step S5 and the power estimated value calculated in the step S6, and predicting the photovoltaic power in the future set time.
2. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 1, wherein the step S2 is characterized in that the data obtained in the step S1 is adopted to construct an irradiance and distributed photovoltaic power station power output mapping model, and the method comprises the following steps:
the irradiation intensity is related to the power output of the distributed photovoltaic power station by adopting the following formula:
wherein,an estimate representing the power produced by a given photovoltaic power plant; t represents the number of time steps in the data;is indicated at +.>A vector of irradiance observed thereat; alpha represents a vector of the inclination angle of the photovoltaic module; beta represents a vector containing the azimuth of the photovoltaic module; g represents a vector containing the geographical coordinates of the power station, including latitude, longitude, altitude;is indicated at +.>A vector of temperatures observed at; p (P) nom A vector representing the nominal power of the photovoltaic module;
different irradiance value functions can return the same power output, and the following steps are adopted for processing:
(1) Estimating a panel orientation from the measured power of a given photovoltaic power plant;
(2) Constructing a function f by using the calculated direction, and solving;
the following formula is used to calculate the optimum value of irradiance:
wherein IRRopt represents an optimized value of irradiance; p represents a photovoltaic power observation;
solving the above formula by using one or several photovoltaic power values is as follows:
wherein n is pv Representing the total number of photovoltaic power signals; p (P) i n Representing the observed value after normalization processing;representing the normalized estimated power value.
3. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 2, wherein the constructing of the map relationship model of the extraterrestrial irradiance and the surface irradiance of the selected region in step S3, dividing weather patterns into a sunny day and a cloudy day, respectively constructing the sunny day map relationship model and the cloudy day map relationship model according to the divided weather patterns, and obtaining irradiance estimation values of set positions in different patterns through the map relationship model specifically comprises:
dividing the inverted irradiance value in the set time into a sunny day or a cloudy day; the dividing method adopts five standards: an average value; a maximum value; the line length of irradiance versus time curve; standard deviation of irradiance change rate; the maximum difference between irradiance variation and clear sky time sequence;
constructing a map model of the extraterrestrial irradiance and the surface irradiance of the selected area; the method comprises the steps of selecting external irradiance, ambient temperature, solar altitude angle and solar azimuth angle as input, taking surface irradiance as output, and respectively constructing a sunny day mapping relation model and a cloudy day mapping relation model according to divided weather patterns through a BP neural network;
the outside radiation illuminance at the actual plant position was calculated using the following formula:
G＝kG 0 sinh＝kG 0 cosθ z ＝kG 0 (sinδsinφ+cosδcosφcosω)
wherein k represents a solar-earth distance correction coefficient; g 0 Representing the solar constant; θ z Representing zenith angle; delta represents declination of the sun; phi represents the local latitude; omega represents the solar time angle;
and mapping the calculated irradiance outside the ground to the estimated irradiance of the actual station, training the obtained mapping relation model of the irradiance outside the ground and the irradiance on the earth surface, and dividing the mapping relation model into a sunny mapping relation model and a cloudy mapping relation model.
4. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 3, wherein the calculating the cross-correlation coefficient of irradiance sequences in step S4 selects an associated power station within a set region that satisfies a set condition, and specifically comprises:
selecting an associated power station meeting a set condition through irradiance space-time association analysis:
the setting condition comprises selecting a plurality of power stations with maximum correlation coefficients according to the set number;
generating an icosahedron triangular grid on a unit sphere, and analyzing irradiance sequence characteristics and space-time association relations among power stations under different weather conditions aiming at simulated photovoltaic power stations; obtaining an associated power station meeting a set condition by calculating pearson correlation coefficients between irradiance sequences, wherein the calculation comprises sequentially recording irradiance values according to a time sequence in the set time according to irradiance values in the set time, and defining the recorded values as irradiance time sequence data; and analyzing and obtaining the correlation degree of irradiance at different places in the region and the delay condition between irradiance sequences through the cross-correlation coefficient.
5. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 4, wherein the calculating the irradiance estimation value of the given associated power station position in step S5 by using the extraterrestrial irradiance and surface irradiance mapping relation model constructed in step S3 specifically comprises:
and mapping the ground irradiance estimation value of the given position by adopting the constructed ground irradiance and ground irradiance mapping model and the calculated ground irradiance of the given associated power station position and the trained neural network model.
6. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 5, wherein the data obtained in step S1 is used in step S6, the irradiance and distributed photovoltaic power station power output mapping model constructed in step S2 is trained, and the power estimation value is calculated, and specifically includes:
the ground irradiance, the ambient temperature, the solar altitude angle, the solar azimuth angle and the historical power are selected as inputs, the actual power is selected as output, and the BP neural network is selected to build a model.
7. The space-time mapping region irradiance estimation and distributed photovoltaic power prediction method according to claim 6, wherein the power station selected in step S4, the irradiance estimation value calculated in step S5, and the power estimation value calculated in step S6 in step S7 are used to construct a space-time double-scale network, and the method specifically comprises:
(7-1) constructing an undirected graph:
and (3) constructing an undirected graph consisting of a plurality of nodes by adopting the distributed photovoltaic power station selected in the step (S4), and adopting the following formula to express:
G t ＝(P t ,ε,W)
wherein G is t A graph showing time t; p (P) t The representation node set is the power set of each distributed power station at the moment t; epsilon represents the collection of edges between nodes; w represents the weight of the adjacency matrix;
(7-2) graph convolution extracts spatial features of data:
the spatial characteristics of the graph convolution extracted data are obtained through Chebyshev polynomial approximation, and the Chebyshev graph convolution formula is as follows:
wherein Θ represents any parameter and is adjusted by initializing assignment and utilizing error back propagation; g represents a graph structure; x represents an input feature; Θ (L) represents a convolution kernel; k represents the size of a graph convolution kernel; θ k Representing polynomial coefficients;a polynomial expansion approximation representing Laplacian;
(7-3) time characteristics of the gated convolution extracted data:
gating convolution is used to capture temporal features in the data; setting the convolution width to K t For extracting K t-1 Time characteristics of (2);
for each node in graph G, timeConvolutions search K of input elements continuously filling in t Neighborhood, so that the time sequence length is shortened by K each time t-1 ；
Length M, with C, to be input to each node i The time domain convolution sequence of each channel is equivalent toY maps to a single output +.>
The above process is represented by the following formula:
wherein Γ represents a convolution kernel; p represents the first half of the causal convolution into which the input is split; the product of Hadamard; sigma (·) represents that the input P controlling the current state is related to the composition structure and dynamic variance in the time series; q represents the second half of the output split representing the causal convolution;
(7-4) time series of spatio-temporal convolution block joint processing graph structure:
adopting a space-time convolution block to jointly process the time sequence of the graph structure, and fusing the characteristics of space and time;
setting the input and output of the space-time convolution block as a three-dimensional tensor;
calculating the input p of the first block using the following formula l Output p l+1 ：
Wherein,a lower temporal kernel representing a first block; reLU (·) represents a true streamline unit function; theta (theta) l A common kernel representing the graph convolution; />An upper temporal kernel representing a first block;
the loss function of the space-time diagram convolution is calculated using the following formula:
wherein,representing the predicted value; w (W) θ Training parameters representing the model; p is p t+1 Representing the true value; p is p j Representing the output map of the spatio-temporal block, j=t+1, …, t-m+1;
(7-5) predicting photovoltaic power for a future set time:
and training the built space-time diagram convolutional neural network according to the obtained irradiance and power data of the given associated power station, taking each distributed power station as a node, calculating the distance between the nodes by using longitude and latitude, and predicting the photovoltaic power in the future set time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311319910.1A CN117350157A (en) | 2023-10-12 | 2023-10-12 | Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311319910.1A CN117350157A (en) | 2023-10-12 | 2023-10-12 | Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117350157A true CN117350157A (en) | 2024-01-05 |
Family
ID=89356914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311319910.1A Pending CN117350157A (en) | 2023-10-12 | 2023-10-12 | Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117350157A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117879047A (en) * | 2024-03-13 | 2024-04-12 | 国网山西省电力公司经济技术研究院 | Optimization method and system for distributed photovoltaic access distribution network |
| CN118300104A (en) * | 2024-06-06 | 2024-07-05 | 国网山西省电力公司太原供电公司 | Distributed photovoltaic power prediction method, system, electronic equipment and storage medium based on graph neural network |
-
2023
- 2023-10-12 CN CN202311319910.1A patent/CN117350157A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117879047A (en) * | 2024-03-13 | 2024-04-12 | 国网山西省电力公司经济技术研究院 | Optimization method and system for distributed photovoltaic access distribution network |
| CN117879047B (en) * | 2024-03-13 | 2024-05-24 | 国网山西省电力公司经济技术研究院 | Optimization method and system for distributed photovoltaic access distribution network |
| CN118300104A (en) * | 2024-06-06 | 2024-07-05 | 国网山西省电力公司太原供电公司 | Distributed photovoltaic power prediction method, system, electronic equipment and storage medium based on graph neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220373984A1 (en) | Hybrid photovoltaic power prediction method and system based on multi-source data fusion | |
| Sun et al. | Estimating the spatial distribution of solar photovoltaic power generation potential on different types of rural rooftops using a deep learning network applied to satellite images | |
| CN117350157A (en) | Space-time mapping region irradiance estimation and distributed photovoltaic power prediction method | |
| CN109344865B (en) | Data fusion method for multiple data sources | |
| Huang et al. | Improvement in artificial neural network-based estimation of grid connected photovoltaic power output | |
| CN111815021B (en) | Photovoltaic power prediction method based on solar radiation climate characteristic identification | |
| Alomari et al. | A predictive model for solar photovoltaic power using the Levenberg-Marquardt and Bayesian regularization algorithms and real-time weather data | |
| JPWO2019155594A1 (en) | Photovoltaic power generation equipment design support device, design support method, design support program, and learned model creation device for design support | |
| CN110781458B (en) | Method for predicting surface solar irradiance based on mixed regression model | |
| CN107437149B (en) | Method and system for determining output of photovoltaic power station | |
| Zhu et al. | GIScience can facilitate the development of solar cities for energy transition | |
| Elhammoudy et al. | Dandelion Optimizer algorithm-based method for accurate photovoltaic model parameter identification | |
| CN117374956A (en) | Short-term prediction method for photovoltaic power generation of comprehensive energy station | |
| WO2021063461A1 (en) | Method for planning a layout of a renewable energy site | |
| CN114492941A (en) | Whole-county photovoltaic prediction method based on cluster division and data enhancement | |
| Dissawa et al. | Cross-correlation based cloud motion estimation for short-term solar irradiation predictions | |
| Saadaoui et al. | Using GIS and photogrammetry for assessing solar photovoltaic potential on Flat Roofs in urban area case of the city of Ben Guerir/Morocco | |
| Muhammed et al. | Building rooftops extraction for solar PV potential estimation using gis-based methods | |
| CN111652126A (en) | Inversion radiation method based on satellite cloud picture | |
| Dewanto et al. | Application of web 3D GIS to display urban model and solar energy analysis using the unmanned aerial vehicle (UAV) data (Case study: National Cheng Kung university buildings) | |
| Carreno et al. | SoDa: An irradiance-based synthetic solar data generation tool | |
| CN106649943B (en) | A kind of evaluation method of building integrated photovoltaic system inclined-plane total radiation | |
| CN116260141B (en) | Reconstruction method and system for power of photovoltaic power station and reconstruction terminal | |
| CN116192005A (en) | Photovoltaic tracking method, system and medium based on micro space-time scale irradiation prediction | |
| Prakash et al. | Remote sensing using drone and machine learning for computation of rooftop solar energy potential |
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 |