CN110751326B - Photovoltaic day-ahead power prediction method and device and storage medium - Google Patents
Photovoltaic day-ahead power prediction method and device and storage medium Download PDFInfo
- Publication number
- CN110751326B CN110751326B CN201910987639.6A CN201910987639A CN110751326B CN 110751326 B CN110751326 B CN 110751326B CN 201910987639 A CN201910987639 A CN 201910987639A CN 110751326 B CN110751326 B CN 110751326B
- Authority
- CN
- China
- Prior art keywords
- algorithm
- prediction
- model
- data
- photovoltaic power
- 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
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 136
- 238000000034 method Methods 0.000 claims abstract description 88
- 238000012549 training Methods 0.000 claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 27
- 238000003062 neural network model Methods 0.000 claims abstract description 22
- 238000012360 testing method Methods 0.000 claims abstract description 21
- 238000003860 storage Methods 0.000 claims abstract description 20
- 238000010219 correlation analysis Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000007781 pre-processing Methods 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 238000013461 design Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 14
- 238000011161 development Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a photovoltaic day-ahead power prediction method, a photovoltaic day-ahead power prediction device and a storage medium, and belongs to the technical field of photovoltaic power prediction. The method comprises the steps of extracting sample data required by training from a historical database according to predicted time; inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm; respectively inputting test data into the SVM model, the engineering coefficient algorithm model and the BP neural network model to obtain a first photovoltaic power predicted value; constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm; and inputting the corresponding photovoltaic power predicted value into the combined prediction algorithm model to obtain a second photovoltaic power predicted value, so that the photovoltaic day-ahead power prediction precision is greatly improved.
Description
Technical Field
The invention belongs to the technical field of photovoltaic power prediction, and particularly relates to a photovoltaic day-ahead power prediction method, a photovoltaic day-ahead power prediction device and a storage medium.
Background
The photovoltaic power prediction refers to the fact that from the known conditions of a power system, economy, society, weather and the like, through analysis and research on historical data, internal connection and development change rules among things are explored, and photovoltaic power development is estimated and presumed in advance. Photovoltaic power prediction is the basic work of departments such as power system planning, power utilization, scheduling and the like. The high-quality photovoltaic power prediction can effectively reduce the power generation cost, ensure that more spot goods are bought on the basis that the power station preferentially completes the basic plan, improve the power generation efficiency of the power station, maximize the income of enterprises in the transaction process and realize the new market target of national energy. Since the photovoltaic power is influenced by nonlinear factors such as temperature and weather conditions and has randomness, the photovoltaic power prediction is a very complex process.
The traditional photovoltaic power prediction method for the power system comprises a unit consumption method, an elastic coefficient method and the like. However, with the development of the national economy, the accuracy of the prediction methods cannot be well guaranteed. Due to the rapid development of the artificial intelligence technology, the grey theory method neural network method and the fuzzy clustering method are widely applied to the field of photovoltaic power prediction. Compared with the prior art, the gray theory method, the neural network method and the fuzzy clustering method can better process the nonlinear relation between the photovoltaic power and the influence factors, thereby obtaining higher prediction precision. In the prior art, photovoltaic power prediction is performed by using a neural network method, wherein a forward neural network can fit a high-dimensional and nonlinear mapping relation between output of a human from complex sample data through learning, so that high-precision prediction is performed. In addition, through DLF fluctuation analysis, photovoltaic power with low fluctuation can be directly distributed or the number of algorithms participating in training is reduced according to system load, so that the calculation time is greatly saved.
Chinese patent publication No. CN109523084A discloses a wind power ultra-short term prediction method based on principal component analysis and machine learning, and relates to the technical field of power systems. Firstly, acquiring n groups of wind power data sample sets, and processing missing values in an interpolation mode; then, performing principal component analysis on an influence factor matrix influencing the wind power in the n groups of sample historical data, and determining transformed principal component data according to the accumulated contribution rate; then combining the principal component data with n groups of wind power ending to t-1 time to serve as input of a machine learning neural network model, outputting a wind power sequence at the t time as a target, performing parameter adjustment training on the model, and storing the trained optimal model; and finally, inputting a prediction sample into the model to obtain a prediction sequence of the wind power.
However, the gray prediction model can be theoretically applied to any photovoltaic power prediction index with nonlinear change, but once the dispersion of data is large, the prediction precision is greatly reduced, and the method is generally not accurate in prediction for a long time interval. Therefore, when the photovoltaic power is predicted by using the neural network method, the photovoltaic power is influenced by various factors, a complex nonlinear phenomenon exists, the variation form of a photovoltaic power curve is greatly different from that of a historical photovoltaic power curve, and the photovoltaic power under various conditions cannot be predicted with high precision by using the forward neural network. And the pure BP neural network needs a large amount of training data, and the data amount in actual production may not meet the training requirement.
Disclosure of Invention
1. Problems to be solved
The photovoltaic day-ahead power prediction method aims at the problem that due to the fact that the dispersion of processed data is large, prediction accuracy is greatly reduced. The invention provides a photovoltaic day-ahead power prediction method, a photovoltaic day-ahead power prediction device and a storage medium.
2. Technical scheme
In order to solve the above problems, the present invention adopts the following technical solutions.
The invention provides a photovoltaic day-ahead power prediction method, which comprises the following steps:
extracting sample data required by training from a historical database according to the predicted time, and preprocessing the sample data;
inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm;
respectively inputting test data into the SVM model, the engineering coefficient algorithm model and the BP neural network model to obtain a first photovoltaic power predicted value;
constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
and inputting the corresponding first photovoltaic power predicted value into the combined prediction algorithm model to obtain a second photovoltaic power predicted value.
Preferably, the second photovoltaic power predicted value of the previous day is classified through the DLF volatility index, and when the volatility is higher than a threshold value, a linear combination prediction algorithm model is selected to process the first photovoltaic power predicted value of the current day; and when the volatility is lower than the threshold value, processing the sample data on the current day by selecting an engineering coefficient algorithm model.
Preferably, the calculation formula of the DLF volatility index is as follows:
wherein: n is expressed as the number of photovoltaic power data points in one day, and load is expressed as the photovoltaic power data vector of a single previous day i The photovoltaic power of the data point of the ith is represented, and DLF represents the volatility indicator.
Preferably, the step of inputting the preprocessed sample data into an SVM algorithm for training to obtain an SVM model includes:
normalizing the preprocessed sample data;
carrying out correlation analysis on the normalized data, and carrying out secondary screening on the sample data;
and inputting the sample data subjected to secondary screening into a kernel function for training to obtain the SVM model.
Preferably, the step of performing correlation analysis on the normalized data includes performing correlation analysis according to the following formula:
where N is the number of photovoltaic power data points for one day, and the sample data input is X = { X = 1 ,x 2 ,x 3 ,L,x i ,L,x N }; the output data is Y = { Y = 1 ,y 2 ,y 3 ,L,y i ,L,y N }; r represents a correlation coefficient.
Preferably, the step of constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm includes:
constructing an objective function when the first photovoltaic power prediction error of the previous day is minimum;
setting the observed value of the first photovoltaic power prediction value object as x t (i＝1,2,K,N)；
Setting m single prediction methods for a first photovoltaic power prediction value object, and setting the prediction value of the ith single prediction method as x it (i =1, 2.. Said., m), specifically, inputting test data into the SVM model to obtain a first photovoltaic power predicted value X 1t (ii) a Inputting test data to the engineering coefficient algorithm model to obtain a first photovoltaic power predicted value X 2t Inputting test data into the BP neural network model to obtain a photovoltaic power predicted value X 3t (ii) a E is then it ＝x t -x it For the prediction error of the ith single prediction method, let l i (i =1,2, \8230;, m) is a weighting coefficient of the ith single prediction method, and should satisfy
Is provided with
Is x t The combined predicted value of (1) is set as e t For the prediction error of the combined prediction, then
While J represents the sum of the squares of the prediction errors of the combined predictions, then
The linear combination prediction model based on the least square prediction error criterion can be obtained as follows:
let L = (L) 1 ,l 2 ,…,l m ) T ，R＝(1,1,…,1) T ，E i ＝(e i1 ,e i2 ,…,e iV ) T Where L denotes a column vector of combined prediction weighting coefficients, R denotes an m-dimensional column vector of elements all 1, E i A column vector of prediction errors representing the ith single prediction method; t represents the transpose of the matrix; reissue to
E＝(E ij ) m×m I.e. the element in E ith row and jth column is E ij (ii) a E is called the combined prediction error information matrix, the original model can be represented in matrix form:
min J＝L T EL
s.t.R T L＝1
using the lagrange multiplier method, one can find:
wherein L is * The representation is the weight (variable value) corresponding to each prediction algorithm; j is a unit of * And the representative combined prediction optimal predicted value vector is the error between the photovoltaic power predicted value of the previous day and the actual value of the current day.
Preferably, the step of preprocessing the sample data comprises:
setting a threshold value, and selecting the sample data which accords with the threshold value range;
and repairing the vacancy value in the sample data by a linear interpolation method.
Preferably, the sample data comprises photovoltaic power historical data and meteorological historical data.
A second aspect of the present invention provides a photovoltaic day-ahead power prediction device, including:
the sample data processing unit is used for extracting sample data required by training from a historical database according to the predicted time and preprocessing the sample data;
the model training unit is used for inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm;
the first calculation unit is used for inputting the SVM model, the engineering coefficient algorithm model and the BP neural network model into test data to obtain a corresponding photovoltaic power predicted value;
the linear combination prediction algorithm model construction unit is used for constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
and the second calculation unit is used for inputting the corresponding photovoltaic power predicted value into the combined prediction algorithm model to obtain an optimal photovoltaic power predicted value.
A third aspect of the invention provides a storage medium storing a computer program comprising program instructions which, when executed by a processor, cause the processor to perform a method as described above.
3. Advantageous effects
Compared with the prior art, the invention has the beneficial effects that:
(1) The method introduces an engineering coefficient method, ensures the robustness of photovoltaic prediction under a low-training sample scene, only needs to collect data of the last 1-7 days by the engineering coefficient method, can directly obtain a more accurate result without training by using the method based on the model of the invention, and ensures the robustness of a prediction algorithm under the condition of serious data loss;
(2) The invention uses a combined prediction method, combines the prediction results of an engineering coefficient method, a BP neural network algorithm and an SVM algorithm in a linear weighting mode, and updates the linearly weighted weights of the three algorithms according to the residual error predicted by the load in the previous week by using the combined prediction algorithm, thereby ensuring the accuracy of the algorithm to the maximum extent;
(3) By introducing DLF volatility index, the invention can avoid training learning of single or all machine learning algorithms under the premise of ensuring prediction precision in a large number of scenes, thereby greatly improving the efficiency of photovoltaic power prediction; through tests, the combined prediction algorithm based on the DLF volatility index has low requirements on computing capacity, and can be operated on a hardware terminal with low configuration.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
fig. 1 is a schematic flow chart a of a photovoltaic day-ahead power prediction method according to an embodiment of the present invention;
fig. 2 is a schematic flow chart b of a photovoltaic power prediction method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a photovoltaic day-ahead power prediction apparatus according to an embodiment of the present invention.
Detailed Description
The basic concept of the invention is as follows: by using a combined prediction method, the prediction results of an engineering coefficient method, a BP neural network algorithm and an SVM algorithm are combined in a linear weighting mode, and the combined prediction algorithm updates the linearly weighted weights of the three algorithms according to the residual error of the previous week load prediction, so that the accuracy of the algorithm is ensured to the maximum extent.
Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and therefore are only examples, and the protection scope of the present invention is not limited thereby. It is to be noted that, unless otherwise specified, technical or scientific terms used herein shall have the ordinary meaning as understood by those skilled in the art to which the invention pertains.
In the present application, the terms "first", "second", etc. are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
As used in this specification and the appended claims, the term "if" may be interpreted contextually as "when", "upon" or "in response to a determination" or "in response to a detection".
In particular implementations, the terminals described in embodiments of the invention include, but are not limited to, other portable devices such as mobile phones, laptop computers, or tablet computers having touch sensitive surfaces (e.g., touch screen displays and/or touch pads). It should also be understood that in some embodiments, the device is not a portable communication device, but is a desktop computer having a touch-sensitive surface (e.g., a touch screen display and/or a touchpad).
In the discussion that follows, a terminal that includes a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, mouse, and/or joystick.
The terminal supports various applications, such as one or more of the following: a drawing application, a presentation application, a word processing application, a website creation application, a disc burning application, a spreadsheet application, a gaming application, a telephone application, a video conferencing application, an email application, an instant messaging application, an exercise support application, a photo management application, a digital camera application, a web browsing application, a digital music player application, and/or a digital video player application.
Various applications that may be executed on the terminal may use at least one common physical user interface device, such as a touch-sensitive surface. One or more functions of the touch-sensitive surface and corresponding information displayed on the terminal can be adjusted and/or changed between applications and/or within respective applications. In this way, a common physical architecture (e.g., touch-sensitive surface) of the terminal can support various applications with user interfaces that are intuitive and transparent to the user.
Example 1
As shown in fig. 1, a photovoltaic day-ahead power prediction method includes:
step S102: extracting sample data required by training from a historical database according to the predicted time, and preprocessing the sample data;
specifically, the types of the sample data may include photovoltaic power historical data, meteorological data and other related factors, the meteorological data includes illumination intensity, rainfall, temperature, humidity, air pressure, wind level and the like, and other related factors include holiday information, weekday information and the like, wherein the photovoltaic power historical data and the illumination intensity are necessary, and other sample data may be selected or supplemented according to actual conditions on site.
For example, the prediction time is set to tomorrow morning 00 to 23; the sample data is extracted from the historical data until the latest record is extracted from the historical database, and the sample data is extracted according to the principle that the closer the sample data is to the prediction time, the larger the extracted data amount is, and the farther the sample data is from the prediction time, the smaller the extracted data amount is, such as the hour load data in the latest month and the day load data on the same date in the year.
Further, preprocessing the extracted data, including setting a threshold value, and selecting the sample data which meets the threshold value range; specifically, all sample data (including bad data) is selected, deleted by setting a threshold range after selection (the position of the data is still, but is a null value), and then the null value (some are originally null values, and some are null values after data are deleted artificially) is supplemented by a linear interpolation method. Here the upper threshold may be taken to be 4 times the average value of the non-zero photovoltaic power at the day and the lower threshold may be taken to be 0.
Step S104: inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm;
(1) For the engineering coefficient method, it should be noted that the historical data before the prediction date is linearly weighted, and the historical data generally involved in the engineering coefficient method operation is data of the previous day, data of the previous three days, or data of the previous week for three days, and the application is mature in the field and is not described herein again.
By introducing the engineering coefficient method, the robustness of photovoltaic prediction under a low-training sample scene is ensured, the engineering coefficient method only needs to collect data of the latest 1-7 days, and by using the method based on the model, a more accurate result can be directly obtained without training, so that the robustness of a prediction algorithm under the condition of serious data loss is ensured.
(2) Inputting the preprocessed sample data into an SVM algorithm for training to obtain an SVM model, wherein the step of obtaining the SVM model comprises the following steps:
(a) Normalizing the preprocessed sample data;
a mapping of non-numeric data is established, for example, for weekday and holiday information, weekday corresponds to 1 and holiday corresponds to 0.
Data normalization, because power, illumination, rainfall and the like are different dimensions, in order to prevent the influence of rounding errors, normalization operation is carried out, and the general operation is to normalize all data to be in a range of [0,1] or [ -1,1 ];
(b) Performing correlation analysis on the normalized data, performing secondary selection on the sample data, and performing correlation analysis to find out data with strong correlation between input and output, so that the difficulty of model training is reduced, and the influence of low-correlation data on the precision and the training speed of the model is avoided;
specifically, correlation analysis is performed on the normalized data, secondary sample selection is performed, and only data with a large correlation coefficient is selected for model training. General correlation analysis is performed using Pearson correlation coefficients, and correlation selection may be performed using other methods. If a certain column of input data of the training set with N samples is X = { X = { (X) 1 ,x 2 ,x 3 ,L,x i ,L,x N Output data of Y = { Y = 1 ,y 2 ,y 3 ,L,y i ,L,y N And f, calculating a Pearson correlation coefficient r according to the following formula:
where N is the number of photovoltaic power data points for one day, and the sample data input is X = { X = 1 ,x 2 ,x 3 ,L,x i ,L,x N }；
The output data is Y = { Y = 1 ,y 2 ,y 3 ,L,y i ,L,y N }; r represents a correlation coefficient.
(c) Inputting the secondary sample data into a kernel function for training to obtain an SVM model;
specifically, a kernel function is selected, the commonly used kernel function includes a polynomial kernel, a gaussian kernel, a linear kernel and the like, and sample data selected twice is input into the kernel function and trained to obtain the SVM model.
(3) Inputting the preprocessed sample data into a BP neural network algorithm to obtain a BP neural network model, wherein the training step comprises the following steps:
(1) normalizing the sample data after pretreatment;
(2) screening sample data according to correlation analysis;
(3) selecting a proper activation function, wherein the commonly used activation functions comprise a Sigmoid function, a Tanh function, a ReLU function and the like;
(4) the method includes inputting sample data to train the BP neural network model, and it should be noted that different from the SVM model, the BP neural network model generally has a larger sample size.
It should be noted that the SVM algorithm, BP algorithm and engineering coefficient method in the combined prediction method in this embodiment may be replaced by other prediction algorithms
Step S106: respectively inputting test data into the SVM model, the engineering coefficient algorithm model and the BP neural network model to obtain a first photovoltaic power predicted value;
step S108: constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
let x be a set of observations of a prediction object (first PV Power predictor object) t (i =1,2,k, n), m single prediction methods exist as the prediction target, and the predicted value of the i-th single prediction method is assumed to be x it (i =1,2,K, m), then e it ＝x t -x it For the prediction error of the ith single prediction method, let l i (i =1,2, \8230;, m) is a weighting coefficient of the ith single prediction method, and should satisfy
In the application scenario of the embodiment, three single prediction algorithms including an SVM algorithm, a BP neural network algorithm and an engineering coefficient method exist, the objective function of the optimization model is that the photovoltaic power prediction error of the previous day is the minimum, and an objective function model is constructed.
Setting a first photovoltaic power prediction value objectObserved value of (1) is x t (i＝1,2,K,N)；
Setting m single prediction methods for a first photovoltaic power predicted value object, and setting the predicted value of the ith single prediction method as x it (i =1, 2.. Said., m), specifically, inputting test data into the SVM model to obtain a first photovoltaic power predicted value X 1t (ii) a Inputting test data into the engineering coefficient algorithm model to obtain a first photovoltaic power predicted value X 2t Inputting test data to the BP neural network model to obtain a photovoltaic power predicted value X 3t (ii) a E is then it ＝x t -x it For the prediction error of the ith single prediction method, let l i (i =1,2, \ 8230;, m) is a weighting coefficient of the ith single prediction method, and should satisfy
Is provided with
Is x t The combined predicted value of (1) is set as t For the prediction error of the combined prediction, then
While J represents the sum of the squares of the prediction errors of the combined predictions, then
From this, a linear combination prediction model based on the least squares of prediction errors as a criterion can be obtained as follows:
let L = (L) 1 ,l 2 ,…,l m ) T ，R＝(1,1,…,1) T ，E i ＝(e i1 ,e i2 ,…,e iV ) T Where L denotes a column vector of combined prediction weighting coefficients, R denotes an m-dimensional column vector of elements all 1, E i A prediction error column vector representing the ith single prediction method; t represents the transpose of the matrix; reissue to
E＝(E ij ) m×m I.e. the element in the ith row and the jth column is E ij (ii) a E is called the combined prediction error information matrix, the original model can be represented in matrix form:
min J＝L T EL
s.t.R T L＝1
using the lagrange multiplier method, one can find:
wherein L is * The representation is the weight (variable value) corresponding to each prediction algorithm; j is a unit of * And the representative combined prediction optimal predicted value vector is the error between the photovoltaic power predicted value of the previous day and the actual value of the current day.
Step S110: and inputting the corresponding photovoltaic power predicted value into the combined prediction algorithm model to obtain a second photovoltaic power predicted value.
And obtaining an optimal prediction result according to the weight (variable value) corresponding to each prediction algorithm, and giving an optimal photovoltaic power prediction curve in the day ahead. Specifically, the first photovoltaic power predicted value X is obtained through the SVM 1t (ii) a Inputting test data into the engineering coefficient algorithm model to obtain a first photovoltaic power predicted value X 2t Inputting test data to the BP neural network model to obtain a photovoltaic power predicted value X 3t Pass through the weight L * The best prediction result is obtained as [ x ] 1t x 2t x 2t ]·L * 。
It should be noted that the algorithm is suitable for combined prediction of more algorithms, and the SVM algorithm, the BP algorithm and the engineering coefficient method in the combined prediction algorithm of the embodiment may be replaced by other prediction algorithms; the number of single prediction algorithms that participate in the combined prediction may vary and still achieve approximately the same result.
As shown in fig. 2, as a variation, the second photovoltaic power predicted value of the previous day is classified by a DLF volatility index, when the volatility is higher than a threshold, the first photovoltaic power predicted value of the current day is processed by using a linear combination prediction algorithm model, and when the volatility is lower than the threshold, the sample data of the current day is processed by using an engineering coefficient algorithm model. The threshold range here may be 7-100, 5-100 or 0-5; those skilled in the art should understand that the scope herein can be obtained by actual data testing, and the system may need to be adjusted for different systems, and is not limited herein.
Classifying the prediction problems according to the DLF volatility index of the previous day, carrying out algorithm prediction on the prediction problems with larger volatility, rapidly distributing the prediction problems with smaller volatility directly according to the system load to obtain a prediction result, combining an engineering coefficient method to form a linear combination prediction method based on prediction error least square,
the calculation formula of the DLF volatility index is as follows:
wherein: n is the number of photovoltaic power data points in one day, load = { load = { (load) 1 ,load 1 ,L load N The photovoltaic power data vector for the day before a certain prediction problem. Note that the DLF fluctuation index employs an absolute value of error | load i+1 -load i The | is used as the measurement of fluctuation, can be replaced by the forms of variance, standard deviation, weighted variance, weighted standard deviation and the like, and can also be selected according to other similar fluctuation indexes to realize the selection of a prediction algorithmThe invention has similar technical steps.
By introducing DLF volatility index, under a large number of application scenes, training learning of a single or all machine learning algorithms can be avoided on the premise of ensuring prediction accuracy, and the efficiency of photovoltaic power prediction is greatly improved; through tests, the combined prediction algorithm based on the DLF volatility index has low requirements on computing capacity, and can be operated on a hardware terminal with low configuration.
Example 2
As shown in fig. 3, the present embodiment provides a photovoltaic day-ahead power prediction apparatus including:
the sample data processing unit 10 is configured to extract sample data required for training from a historical database according to the predicted time, and preprocess the sample data;
the model training unit 20 is used for inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm;
the first calculating unit 30 is configured to input the SVM model, the engineering coefficient algorithm model and the BP neural network model into test data to obtain a corresponding photovoltaic power predicted value;
the linear combination prediction algorithm model building unit 40 is used for building a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
and the second calculating unit 50 is configured to input the corresponding photovoltaic power predicted value into the combined prediction algorithm model to obtain an optimal photovoltaic power predicted value.
Example 3
The present embodiment provides a computer-readable storage medium storing a computer program comprising program instructions that, when executed by a processor, cause the processor to perform the method of embodiment 1.
Specifically, the computer-readable storage medium may be an internal storage unit of the terminal according to the foregoing embodiment, for example, a hard disk or a memory of the terminal. The computer readable storage medium may also be an external storage device of the terminal, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like provided on the terminal. Further, the computer-readable storage medium may also include both an internal storage unit and an external storage device of the terminal. The computer-readable storage medium is used for storing the computer program and other programs and data required by the terminal. The computer readable storage medium may also be used to temporarily store data that has been output or is to be output.
Those of ordinary skill in the art will appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the components and steps of the various examples have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the terminal and the unit described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed terminal and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electrical, mechanical or other form of connection.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one position, or may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit may be implemented in the form of hardware, or may also be implemented in the form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially or partially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit and scope of the present invention, and they should be construed as being included in the following claims and description.
Claims (6)
1. A photovoltaic day-ahead power prediction method is characterized by comprising the following steps:
extracting sample data required by training from a historical database according to the predicted time, and preprocessing the sample data;
inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm; the engineering coefficient algorithm is used for carrying out linear weighting on historical data before the prediction date;
respectively inputting test data into the SVM model, the BP neural network model and the engineering coefficient algorithm model to obtain a corresponding first photovoltaic power predicted value;
constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
inputting the corresponding first photovoltaic power predicted value into the combined prediction algorithm model to obtain a second photovoltaic power predicted value;
classifying a second photovoltaic power predicted value of the previous day through a DLF volatility index, and when the volatility is higher than a threshold value, selecting a linear combination prediction algorithm model to process a first photovoltaic power predicted value of the current day; when the volatility is lower than the threshold value, processing the sample data of the current day by selecting an engineering coefficient algorithm model;
the step of inputting the preprocessed sample data into an SVM algorithm for training to obtain an SVM model comprises the following steps:
normalizing the preprocessed sample data;
carrying out correlation analysis on the normalized data, and carrying out secondary screening on the sample data;
inputting the sample data subjected to secondary screening into a kernel function for training to obtain an SVM model;
performing correlation analysis on the normalized data, performing secondary sample selection, and performing model training by selecting only data with a large correlation coefficient; a general correlation analysis is performed using correlation coefficients, if a certain column of input data of the training set with N samples is X = { X = { (X) } 1 ,x 2 ,x 3 ,…,x s ,…,x N Output data of Y = { Y = } 1 ,y 2 ,y 3 ,…,y s ,…,y N And f, calculating a Pearson correlation coefficient r according to the following formula:
wherein N represents the number of photovoltaic power data points in one day; r represents a correlation coefficient;
the calculation formula of the DLF volatility index is as follows:
wherein: n is expressed as the number of photovoltaic power data points in one day, and load is expressed as the photovoltaic power data vector of a certain previous day s The photovoltaic power at the s-th data point is shown, and the DLF is a volatility indicator.
2. The photovoltaic day-ahead power prediction method according to claim 1, characterized in that: the step of constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm comprises the following steps:
constructing an objective function when the first photovoltaic power prediction error of the previous day is minimum;
setting the observed value of the first photovoltaic power prediction value object as x t ，t＝1,2,…,N；
Set to be firstThe photovoltaic power predicted value object has m single prediction methods, and the predicted value of the ith single prediction method is set as x it I =1,2, \8230m, specifically, test data is input into the SVM model to obtain a first photovoltaic power predicted value X 1t (ii) a Inputting test data into the engineering coefficient algorithm model to obtain a first photovoltaic power predicted value X 2t (ii) a Inputting test data into the BP neural network model to obtain a photovoltaic power predicted value X 3t (ii) a Then e it ＝x t -x it For the prediction error of the ith single prediction method, let l i I =1,2, \ 8230, m is the weighting coefficient of the ith single prediction method and should satisfy
Is provided with
Is x t The combined predicted value of (1) is set as t For the prediction error of the combined prediction, there are
While J represents the sum of the squares of the prediction errors of the combined predictions, then
From this, a linear combination prediction model based on the least squares of prediction errors as a criterion can be obtained as follows:
let L = (L) 1 ,l 2 ,…,l m ) T ，R＝(1,1,…,1) T ，E i ＝(e i1 ,e i2 ,…,e iV ) T Where L denotes a column vector of combined prediction weighting coefficients, R denotes an m-dimensional column vector of elements all 1, E i A vector representing a prediction error column v of the i-th single prediction method; t represents the transpose of the matrix; reissue to order
E＝(E ij ) m×m I.e. E row i column j element is
E is called the combined prediction error information matrix, the original model can be represented in matrix form:
min J＝L T EL
s.t.R T L＝1
by using the Lagrange multiplier method, the following can be obtained:
wherein L is * The representativeness is the weight corresponding to each prediction algorithm; j. the design is a square * And the representative combined prediction optimal predicted value vector is the error between the photovoltaic power predicted value of the previous day and the actual value of the current day.
3. The photovoltaic day-ahead power prediction method according to claim 1, characterized in that: the step of preprocessing the sample data comprises:
setting a threshold value, and selecting the sample data which accords with the threshold value range;
and repairing the vacancy value in the sample data by a linear interpolation method.
4. The photovoltaic day-ahead power prediction method according to any one of claims 1 or 3, characterized in that: the sample data includes photovoltaic power historical data and meteorological historical data.
5. A photovoltaic day-ahead power prediction apparatus, comprising:
the sample data processing unit is used for extracting sample data required by training from a historical database according to the predicted time and preprocessing the sample data;
the model training unit is used for inputting the preprocessed sample data into an SVM algorithm and a BP neural network algorithm for training to obtain an SVM model and a BP neural network model; constructing an engineering coefficient algorithm model according to an engineering coefficient algorithm; the engineering coefficient algorithm is used for linearly weighting historical data before a prediction day;
the first calculation unit is used for respectively inputting test data into the SVM model, the engineering coefficient algorithm model and the BP neural network model to obtain a corresponding first photovoltaic power predicted value;
the linear combination prediction algorithm model construction unit is used for constructing a linear combination prediction algorithm model according to the SVM algorithm, the BP neural network algorithm and the engineering coefficient algorithm;
the second calculation unit is used for inputting the corresponding first photovoltaic power predicted value into the combined prediction algorithm model to obtain a second photovoltaic power predicted value;
classifying the second photovoltaic power predicted value of the previous day through the DLF volatility index, and when the volatility is higher than a threshold value, selecting a linear combination prediction algorithm model to process the first photovoltaic power predicted value of the current day; when the volatility is lower than the threshold value, processing the sample data of the current day by selecting an engineering coefficient algorithm model;
the step of inputting the preprocessed sample data into an SVM algorithm for training to obtain an SVM model comprises the following steps:
normalizing the preprocessed sample data;
carrying out correlation analysis on the normalized data, and carrying out secondary screening on the sample data;
inputting the sample data subjected to secondary screening into a kernel function for training to obtain an SVM model;
performing correlation analysis on the normalized data, performing secondary sample selection, and performing model training by selecting only data with a large correlation coefficient; a general correlation analysis is performed using correlation coefficients, if a certain column of input data of the training set with N samples is X = { X = { (X) } 1 ,x 2 ,x 3 ,…,x s ,…,x N Output data of Y = { Y = } 1 ,y 2 ,y 3 ,…,y s ,…,y N And f, calculating a Pearson correlation coefficient r according to the following formula:
wherein N represents the number of photovoltaic power data points in one day; r represents a correlation coefficient;
the calculation formula of the DLF volatility index is as follows:
wherein: n is expressed as the number of photovoltaic power data points in one day, and load is expressed as the photovoltaic power data vector of a certain previous day s The photovoltaic power at the s-th data point is shown, and the DLF is a volatility indicator.
6. A storage medium, characterized in that the storage medium stores a computer program comprising program instructions which, when executed by a processor, cause the processor to carry out the method according to any one of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910987639.6A CN110751326B (en) | 2019-10-17 | 2019-10-17 | Photovoltaic day-ahead power prediction method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910987639.6A CN110751326B (en) | 2019-10-17 | 2019-10-17 | Photovoltaic day-ahead power prediction method and device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110751326A CN110751326A (en) | 2020-02-04 |
| CN110751326B true CN110751326B (en) | 2022-10-28 |
Family
ID=69278626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910987639.6A Active CN110751326B (en) | 2019-10-17 | 2019-10-17 | Photovoltaic day-ahead power prediction method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110751326B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931981B (en) * | 2020-07-06 | 2022-05-10 | 安徽天尚清洁能源科技有限公司 | Photovoltaic power generation ultra-short-term prediction method based on machine learning multi-model combination |
| CN112565378A (en) * | 2020-11-30 | 2021-03-26 | 中国科学院深圳先进技术研究院 | Cloud native resource dynamic prediction method and device, computer equipment and storage medium |
| CN114764741A (en) * | 2021-01-15 | 2022-07-19 | 深圳光耀智微科技有限公司 | Method and system for predicting running wind power of wind driven generator |
| CN114298141A (en) * | 2021-11-19 | 2022-04-08 | 中国华能集团清洁能源技术研究院有限公司 | Anomaly detection method and system for power generation equipment based on deep learning |
| CN113988491B (en) * | 2021-12-29 | 2022-04-01 | 北京东润环能科技股份有限公司 | Photovoltaic short-term power prediction method and device, electronic equipment and storage medium |
| CN114548804B (en) * | 2022-02-28 | 2024-05-31 | 国网河南省电力公司电力科学研究院 | Day-ahead power prediction method and system based on improved boosting tree |
| CN115587642B (en) * | 2022-06-22 | 2023-07-11 | 大唐海南能源开发有限公司 | BP neural network-based photovoltaic system fault alarm method |
| CN117175585B (en) * | 2023-11-02 | 2024-03-08 | 深圳航天科创泛在电气有限公司 | Wind power power prediction method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105426956A (en) * | 2015-11-06 | 2016-03-23 | 国家电网公司 | Ultra-short-period photovoltaic prediction method |
| CN106228278A (en) * | 2016-08-19 | 2016-12-14 | 长园深瑞继保自动化有限公司 | Photovoltaic power prognoses system |
| CN109871976A (en) * | 2018-12-20 | 2019-06-11 | 浙江工业大学 | A power quality prediction method based on clustering and neural network for distribution network with distributed generation |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103218674A (en) * | 2013-04-07 | 2013-07-24 | 国家电网公司 | Method for predicating output power of photovoltaic power generation system based on BP (Back Propagation) neural network model |
| WO2015152941A1 (en) * | 2014-04-04 | 2015-10-08 | Siemens Aktiengesellschaft | Combing multiple trending models for photovoltaic plant output forecasting |
| CN105023070A (en) * | 2015-08-12 | 2015-11-04 | 河海大学常州校区 | Output power prediction method of photovoltaic system |
| US10254641B2 (en) * | 2016-12-01 | 2019-04-09 | Lam Research Corporation | Layout pattern proximity correction through fast edge placement error prediction |
| US11545830B2 (en) * | 2017-01-18 | 2023-01-03 | Board Of Regents, The University Of Texas System | Systems and methods of hierarchical forecasting of solar photovoltaic energy production |
| CN107403015B (en) * | 2017-08-07 | 2020-12-29 | 南京金水尚阳信息技术有限公司 | Short-term optical power prediction method based on time series similarity |
| CN109190786A (en) * | 2018-07-06 | 2019-01-11 | 上海电机学院 | A kind of combination forecasting method of photovoltaic efficiency |
| CN109165774A (en) * | 2018-08-07 | 2019-01-08 | 燕山大学 | A kind of short-term photovoltaic power prediction technique |
| CN109685257A (en) * | 2018-12-13 | 2019-04-26 | 国网青海省电力公司 | A kind of photovoltaic power generation power prediction method based on Support vector regression |
| CN109657881A (en) * | 2019-01-14 | 2019-04-19 | 南京国电南自电网自动化有限公司 | A kind of neural network photovoltaic power generation prediction technique and system suitable for small sample |
-
2019
- 2019-10-17 CN CN201910987639.6A patent/CN110751326B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105426956A (en) * | 2015-11-06 | 2016-03-23 | 国家电网公司 | Ultra-short-period photovoltaic prediction method |
| CN106228278A (en) * | 2016-08-19 | 2016-12-14 | 长园深瑞继保自动化有限公司 | Photovoltaic power prognoses system |
| CN109871976A (en) * | 2018-12-20 | 2019-06-11 | 浙江工业大学 | A power quality prediction method based on clustering and neural network for distribution network with distributed generation |
Non-Patent Citations (2)
| Title |
|---|
| 光伏发电功率的预测方法研究;王鸿玺;《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》;20140215(第2期);全文 * |
| 大型并网光伏电站功率预测研究;党文强;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20180415(第4期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110751326A (en) | 2020-02-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110751326B (en) | Photovoltaic day-ahead power prediction method and device and storage medium | |
| CN110298501B (en) | Electrical load prediction method based on long-time and short-time memory neural network | |
| CN113177366B (en) | An integrated energy system planning method, device and terminal equipment | |
| CN110751416A (en) | Method, device and equipment for predicting water consumption | |
| CN114330935B (en) | New energy power prediction method and system based on multiple combination strategies integrated learning | |
| CN110796513A (en) | Multitask learning method and device, electronic equipment and storage medium | |
| CN110175689A (en) | A kind of method of probabilistic forecasting, the method and device of model training | |
| CN111506876A (en) | Data prediction analysis method, system, equipment and readable storage medium | |
| CN114118570A (en) | Service data prediction method and device, electronic equipment and storage medium | |
| CN107679680A (en) | A kind of financial forward prediction method, apparatus, equipment and storage medium | |
| CN117909888A (en) | Intelligent artificial intelligence climate prediction method | |
| Zhao et al. | Risk reduction and efficiency increase in large portfolios: Gross-exposure constraints and shrinkage of the covariance matrix | |
| CN115018119A (en) | Power load prediction method and system | |
| CN110597990A (en) | Financial analysis method and system based on intelligent classification | |
| Chen et al. | Assessment of tropical cyclone disaster loss in Guangdong Province based on combined model | |
| Ma et al. | Short-Term Household Load Forecasting Based on Attention Mechanism and CNN-ICPSO-LSTM. | |
| CN118568202A (en) | Query statement generation model processing method and device and computer equipment | |
| CN116562359B (en) | CTR prediction model training method and device based on contrast learning and electronic equipment | |
| Zhao et al. | Forecasting short-term oil price with a generalised pattern matching model based on empirical genetic algorithm | |
| Zhang | Housing price prediction using machine learning algorithm | |
| CN116681467A (en) | Sales predicting method based on improved WOA algorithm and related equipment thereof | |
| CN116257758A (en) | Model training method, crowd expanding method, medium, device and computing equipment | |
| Fan et al. | An ISSA-TCN short-term urban power load forecasting model with error factor | |
| Hong et al. | The study on affecting factors of regional marine industrial structure upgrading | |
| CN115409607A (en) | Method and device for determining credit granting data and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |