CN112949936A - Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron - Google Patents
Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron Download PDFInfo
- Publication number
- CN112949936A CN112949936A CN202110331087.0A CN202110331087A CN112949936A CN 112949936 A CN112949936 A CN 112949936A CN 202110331087 A CN202110331087 A CN 202110331087A CN 112949936 A CN112949936 A CN 112949936A
- Authority
- CN
- China
- Prior art keywords
- day
- similar
- data
- power
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000009466 transformation Effects 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims description 33
- 230000006870 function Effects 0.000 claims description 16
- 230000004913 activation Effects 0.000 claims description 9
- 230000003044 adaptive effect Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 8
- 238000012886 linear function Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000011160 research Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 7
- 238000007619 statistical method Methods 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000000053 physical method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002349 favourable effect Effects 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 210000002966 serum Anatomy 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000005309 stochastic process Methods 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 230000007704 transition Effects 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/148—Wavelet transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Computational Mathematics (AREA)
- Marketing (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Water Supply & Treatment (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Primary Health Care (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Public Health (AREA)
- Algebra (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
Abstract
The invention provides a short-term photovoltaic power prediction method based on similar day wavelet transformation and a multilayer perceptron. The model provided by the invention combines the advantages of various algorithms, and can effectively improve the accuracy of prediction. The main innovation point is that similar-day wavelet transformation and a multilayer perceptron are combined for research of predicting output power of a photovoltaic power station.
Description
Technical Field
The invention belongs to the technical field of short-term prediction of power of photovoltaic power stations, and particularly relates to a short-term photovoltaic power prediction method based on similar-day wavelet transformation and a multilayer perceptron.
Background
In recent years, solar energy has received much attention due to its clean, pollution-free, abundant and inexhaustible characteristics. According to the report issued by the International Energy Agency (IEA), the installed capacity of the global photovoltaic system is at least 627GW by 2019, wherein 115GW in 2019. However, uncertainty, intermittency, fluctuation and uncontrollable property of solar energy are not beneficial to safe and stable operation of a power grid, and large-scale application of photovoltaic power generation is hindered. The accurate prediction of the photovoltaic power is the premise of large-scale application of a photovoltaic system and reasonable planning of a power grid, and the power grid can be configured in advance by predicting the photovoltaic power generation amount, so that the influence of solar energy fluctuation on the power grid is reduced.
Photovoltaic power prediction methods can be divided into two categories according to the principles used: physical methods and statistical methods. The physical method first establishes a physical model of a power plant according to a physical formula, and then substitutes meteorological data obtained through Numerical Weather Prediction (NWP) into the model to predict power of the power plant, which does not require historical data and can obtain a predicted value of power before actual construction of the power plant, but requires many factors to be considered in the model construction process, such as: parameters of the power plant equipment, aging of the power plant equipment, influence of an external environment and the like make modeling difficult, and the anti-interference capability and robustness of the model are poor. The statistical method is based on historical data of the photovoltaic power station, a prediction model is established through correlation analysis, parameter estimation and curve fitting, clear and complete understanding of a complex photoelectric conversion process of a photovoltaic system is not needed, and compared with a physical method, the statistical method has the advantages of simplicity in modeling and high universality.
Conventional statistical methods commonly used techniques include time series methods, fuzzy logic methods, and markov chain methods. Time series analysis is based on time series data obtained by system observation, a mathematical model is established through curve fitting and parameter estimation (such as nonlinear least squares), characteristics and development rules of variables can be found from the time series, and therefore future trends of the variables can be effectively predicted, but prediction results are degraded as time scales and output dimensions are increased. Fuzzy logic simulates the uncertainty concept judgment and reasoning thinking mode of human brain, realizes fuzzy comprehensive judgment, solves the problem of conventional fuzzy information which is difficult to process by the traditional method, and does not need to establish an accurate mathematical model of a research object, so that the model is simple and easy to accept, understand and apply, but establishing fuzzy rules and membership functions for complex systems is very difficult and time-consuming. The markov chain is a typical stochastic process, the current state of the model is only related to the state of the previous moment, and the markov chain predicts the state of the system and the development trend thereof by using a transition probability matrix, and although the markov chain can well predict the state of the system in some cases, the markov chain is not suitable for medium-long term prediction.
The novel statistical method based on the artificial intelligence algorithm can effectively overcome the defects of the traditional statistical method. The artificial intelligence model can extract high-dimensional complex nonlinear characteristics and directly map the characteristics to output, and compared with a physical model and a traditional statistical model, the artificial intelligence model has great improvement in the aspects of prediction accuracy, stability, universality and the like.
In some studies, the similar daily approach is considered as an effective preprocessing step to improve the accuracy of short-term power prediction; the wavelet transform has the capacity of decomposing signals, can decompose the signals into steady-state components and non-steady-state components, and then respectively establishes a prediction model aiming at different components, so that the wavelet transform is favorable for learning the overall change trend of the signals without losing tiny detailed signals; in addition, the neural network used is a multilayer perceptron which has strong nonlinear fitting capability.
Disclosure of Invention
Aiming at the problems in the prior art, the invention aims to provide a short-term photovoltaic power prediction method based on similar day wavelet transformation and a multilayer perceptron. The model provided by the invention combines the advantages of various algorithms, and can effectively improve the accuracy of prediction. The main innovation point is that similar-day wavelet transformation and a multilayer perceptron are combined for research of predicting output power of a photovoltaic power station.
Which comprises the following steps: step S1: analyzing meteorological parameters influencing photovoltaic power by using the correlation coefficient, and finally selecting the four most relevant meteorological parameters as the input of the model; step S2: processing 20 days history data before the day to be predicted, eliminating abnormal values and values of the night, then selecting similar day data, and carrying out normalization processing on historical power and historical meteorological parameters to obtain a training data set; step S3: decomposing historical power data and historical meteorological data in a training data set into a steady-state component and an unsteady-state component by using wavelet transformation, wherein the number of the unsteady-state components can be multiple; step S4: the multi-layer perceptron is adopted to learn the training data set after wavelet transformation, and a multi-layer perceptron model is respectively trained on the steady-state component and the unsteady-state component, namely the power of the steady-state part is predicted by the meteorological parameters of the steady-state part, the power of the unsteady-state part is predicted by the meteorological parameters of the unsteady-state part, and the parameters of the network are adjusted by an adaptive moment estimation algorithm; step S5: normalizing the meteorological parameters of the day to be predicted and decomposing the meteorological parameters into a steady-state component and an unsteady-state component by using wavelet transformation, then taking the steady-state component and the unsteady-state component as the input of a corresponding model to obtain each predicted power value component, and summing and inversely normalizing each power component to obtain a predicted final power value; the technical scheme of the invention can obviously improve the accuracy and reliability of short-term power prediction of the photovoltaic power station.
The invention specifically adopts the following technical scheme:
a short-term photovoltaic power prediction method based on similar-day wavelet transform and a multilayer perceptron is characterized in that:
step S1: analyzing meteorological parameters influencing photovoltaic power by using the correlation coefficients, and selecting various meteorological parameters as the input of a model;
step S2: processing historical data of a plurality of days before the day to be predicted, eliminating abnormal values and values of the night, then selecting similar day data, and carrying out normalization processing on historical power and historical meteorological parameters to obtain a training data set;
step S3: decomposing historical power data and historical meteorological parameters in a training data set into a steady-state component and an unsteady-state component by using wavelet transformation, wherein the number of the unsteady-state components is one or more;
step S4: and (3) learning the training data set after wavelet transformation by adopting a multilayer perceptron, and respectively training a multilayer perceptron model for the steady-state component and the unsteady-state component: the power of the steady-state part is predicted by the meteorological parameters of the steady-state part, the power of the unsteady-state part is predicted by the meteorological parameters of the unsteady-state part, and the parameters of the network are adjusted by an adaptive moment estimation algorithm;
step S5: the meteorological parameters of the day to be predicted are normalized and decomposed into a steady-state component and an unsteady-state component by using wavelet transformation, then the steady-state component and the unsteady-state component are used as the input of a corresponding model to obtain each predicted power value component, and the power components are summed and inversely normalized to obtain the predicted final power value.
Preferably, the meteorological parameters adopted in step S1 include: global level irradiance, scatter level irradiance, ambient temperature, and relative humidity.
Preferably, in step S2, the historical data 20 days before the day to be predicted is processed, the data corresponding to the part of the power variation caused by non-meteorological parameters (such as system faults and human factors) is removed, the data at night is removed, the data from day 6 to 19 are taken, the selection of similar days is performed by calculating the euclidean distance of the global level irradiance between the test set and the training set, and the first 80% data with the minimum euclidean distance in the training set is selected as a new training set and normalized to the interval [0,1 ].
Preferably, in step S3, the basis function of the wavelet transform is bior6.8 in the biocathogonal wavelet series.
Preferably, step S4 specifically includes the following steps:
step S41: forward propagation: a multilayer perceptron is adopted, an implicit layer is assumed to be an N layer, and the method comprises the following steps: h1＝f(W1X+B1)，Hi＝f(WiHi-1+Bi),i＝2,3,…,N，O＝g(WoHN+Bo) (ii) a Wherein, W is a weight; b is an offset value; x is the input of the multi-layer perceptron; hiI is 1,2, …, N is the output of the ith hidden layer; o is the output of the multilayer perceptron; f (-) is the activation function of the hidden layer, expressed as
g (-) is the activation function of the output layer, the output layer will select different activation functions during training and prediction, linear function during training is selected, and positive linear function is used during prediction, the expression of which is
Step S42, back propagation: the loss function of the model is mean square error MSE, an adaptive moment estimation algorithm Adam is used, and pseudo codes of the adaptive moment estimation algorithm Adam are as follows:
compared with the prior art, the invention and the preferred scheme thereof have the following advantages: the adopted similar daily method is an effective preprocessing step for improving the short-term power prediction accuracy; the wavelet transform has the capacity of decomposing signals, can decompose the signals into steady-state components and non-steady-state components, and then respectively establishes a prediction model aiming at different components, so that the wavelet transform is favorable for learning the overall change trend of the signals without losing tiny detailed signals; in addition, the neural network used is a multilayer perceptron which has strong nonlinear fitting capability. The model provided by the scheme combines the advantages of various algorithms, and the result shows that compared with the conventional photovoltaic power station short-term power prediction method, the photovoltaic power station short-term power prediction method has the advantages that the accuracy and the reliability of the photovoltaic power station short-term power prediction are greatly improved.
Drawings
The invention is described in further detail below with reference to the following figures and detailed description:
FIG. 1 is a schematic flow chart of a short-term photovoltaic power prediction method based on similar-day wavelet transform and a multi-layer perceptron according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a short-term photovoltaic power prediction model based on similar-day wavelet transform and a multi-layer perceptron according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a multi-layered sensor according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a prediction result of a short-term photovoltaic power prediction model based on similar-day wavelet transform and a multi-layer perceptron in spring according to an embodiment of the invention;
FIG. 5 is a schematic diagram of a prediction result of a short-term photovoltaic power prediction model based on similar-day wavelet transform and a multi-layer perceptron in summer according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a prediction result of a short-term photovoltaic power prediction model based on similar-day wavelet transform and a multi-layer perceptron in autumn according to an embodiment of the present invention;
fig. 7 is a schematic diagram of a prediction result of a short-term photovoltaic power prediction model based on similar-day wavelet transform and a multi-layer perceptron in winter according to an embodiment of the present invention.
Detailed Description
In order to make the features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail as follows:
the embodiment provides a short-term photovoltaic power prediction method based on similar-day wavelet transform and a multi-layer perceptron, a specific flow diagram of which is shown in fig. 1, and the method comprises the following steps:
step S1: analyzing meteorological parameters influencing photovoltaic power by using the correlation coefficient, and finally selecting the four most relevant meteorological parameters as the input of the model;
step S2: processing 20 days history data before the day to be predicted, eliminating abnormal values and values of the night, then selecting similar day data, and carrying out normalization processing on historical power and historical meteorological parameters to obtain a training data set;
step S3: decomposing historical power data and historical meteorological data in a training data set into a steady-state component and an unsteady-state component by using wavelet transformation, wherein the number of the unsteady-state components can be multiple;
step S4: the multi-layer perceptron is adopted to learn the training data set after wavelet transformation, and a multi-layer perceptron model is respectively trained on the steady-state component and the unsteady-state component, namely the power of the steady-state part is predicted by the meteorological parameters of the steady-state part, the power of the unsteady-state part is predicted by the meteorological parameters of the unsteady-state part, and the parameters of the network are adjusted by an adaptive moment estimation algorithm;
step S5: the meteorological parameters of the day to be predicted are normalized and decomposed into a steady-state component and an unsteady-state component by using wavelet transformation, then the steady-state component and the unsteady-state component are used as the input of a corresponding model to obtain each predicted power value component, and the power components are summed and inversely normalized to obtain the predicted final power value.
In one embodiment of the present invention, in step S1, meteorological parameters are selected as inputs to the model by:
analyzing meteorological parameters influencing photovoltaic power by using the correlation coefficient, and finally selecting four most relevant meteorological parameters as the input of the model, wherein the four finally determined meteorological parameters are as follows: global level irradiance, scatter level irradiance, ambient temperature, and relative humidity.
In an embodiment of the present invention, in step S2, the data is filtered to generate a training data set as follows:
and (3) eliminating data corresponding to the part of power change caused by non-meteorological parameters (such as system faults and human factors), and eliminating data in the dark, wherein data from day 6 to day 19 are taken, similar days are selected by calculating the Euclidean distance of global level irradiance between the test set and the training set, and the data of the first 80% with the minimum Euclidean distance in the training set is selected as a new training set and is normalized to the interval [0,1 ].
In one embodiment of the present invention, in step S3, the historical power data and the historical meteorological data are decomposed into stationary components and non-stationary components by:
decomposing historical power data and historical meteorological data in a training data set into a stable component and an unstable component by using wavelet transformation, wherein the number of the unstable components can be multiple, and a used wavelet transformation basis function is bior6.8 in a Biorthogonal wavelet system;
in an embodiment of the present invention, in step S4, a photovoltaic plant short-term power prediction model map based on similar-day wavelet transform and a multi-layer perceptron is shown in fig. 2, and the multi-layer perceptron model is constructed and network parameters are adjusted as follows:
step S41, forward propagation, using a multi-layer perceptron, assuming that its hidden layer is N layers, the method is: h1＝f(W1X+B1)，Hi＝f(WiHi-1+Bi),i＝2,3,…,N，O＝g(WoHN+Bo) (ii) a Wherein, W is a weight; b is an offset value; x is the input of the multi-layer perceptron; hiI is 1,2, …, N is the output of the ith hidden layer; o is the output of the multilayer perceptron; f (-) is the activation function of the hidden layer, expressed as
g (-) is the activation function of the output layer, the output layer will select different activation functions during training and prediction, linear function during training is selected, and positive linear function is used during prediction, the expression of which is
The structure of the multilayer perceptron is shown in FIG. 3;
step S42, backward propagation, wherein the loss function of the model is mean square error MSE, and an adaptive moment estimation algorithm (Adam) is used, and pseudo codes thereof are as follows:
preferably, in this example, a photovoltaic power prediction is performed under four seasons, namely spring, summer, autumn and winter, with a photovoltaic power station No. 22 (16.8 kW capacity) of DKA Solar center (serum knowledgebase Solar center) in Australia as a research object, and the test data of each season is selected as follows: spring selection 2018/09/29; selecting 2018/02/08 in summer; selecting 2018/05/21 in autumn; selecting 2018/08/21 in winter; the training data was selected as 20 days history data prior to the test data. The predicted effect maps are shown in fig. 4 to 7. As can be seen from the Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE) in table 1, the method provided in this embodiment can perform more accurate prediction, and fully embodies the accuracy of the solution of the present invention in terms of effect.
TABLE 1 Performance testing under four seasonal conditions
| Performance index | Spring | Summer | Autumn | In winter |
| RMSE/kW | 0.63 | 0.79 | 0.57 | 0.11 |
| MAPE/％ | 2.07 | 2.81 | 2.76 | 0.44 |
The present invention is not limited to the above preferred embodiments, and all other various methods for short-term photovoltaic power prediction based on wavelet transform of similar days and multi-layer perceptron can be derived from the present patent, and all equivalent changes and modifications made according to the claimed invention shall fall within the scope of the present patent.
Claims (5)
1. A short-term photovoltaic power prediction method based on similar-day wavelet transform and a multilayer perceptron is characterized in that:
step S1: analyzing meteorological parameters influencing photovoltaic power by using the correlation coefficients, and selecting various meteorological parameters as the input of a model;
step S2: processing historical data of a plurality of days before the day to be predicted, eliminating abnormal values and values of the night, then selecting similar day data, and carrying out normalization processing on historical power and historical meteorological parameters to obtain a training data set;
step S3: decomposing historical power data and historical meteorological parameters in a training data set into a steady-state component and an unsteady-state component by using wavelet transformation, wherein the number of the unsteady-state components is one or more;
step S4: and (3) learning the training data set after wavelet transformation by adopting a multilayer perceptron, and respectively training a multilayer perceptron model for the steady-state component and the unsteady-state component: the power of the steady-state part is predicted by the meteorological parameters of the steady-state part, the power of the unsteady-state part is predicted by the meteorological parameters of the unsteady-state part, and the parameters of the network are adjusted by an adaptive moment estimation algorithm;
step S5: the meteorological parameters of the day to be predicted are normalized and decomposed into a steady-state component and an unsteady-state component by using wavelet transformation, then the steady-state component and the unsteady-state component are used as the input of a corresponding model to obtain each predicted power value component, and the power components are summed and inversely normalized to obtain the predicted final power value.
2. The method for short-term photovoltaic power prediction based on similar-day wavelet transform and multi-layer perceptron as claimed in claim 1, characterized in that: the meteorological parameters used in step S1 include: global level irradiance, scatter level irradiance, ambient temperature, and relative humidity.
3. The method for short-term photovoltaic power prediction based on similar-day wavelet transform and multi-layer perceptron as claimed in claim 1, characterized in that: in step S2, the historical data 20 days before the day to be predicted is processed, the data corresponding to the part of power variation caused by non-meteorological parameters is removed, the data at night is removed, the data from day 6 to 19 are taken, the selection of similar days is performed by calculating the euclidean distance of the global level irradiance between the test set and the training set, the first 80% of the data with the minimum euclidean distance in the training set is selected as a new training set, and the new training set is normalized to the interval [0,1 ].
4. The method for short-term photovoltaic power prediction based on similar-day wavelet transform and multi-layer perceptron as claimed in claim 1, characterized in that: in step S3, the basis function of the wavelet transform is bior6.8 in the biocathogonal wavelet series.
5. The method for short-term photovoltaic power prediction based on similar-day wavelet transform and multi-layer perceptron as claimed in claim 1, characterized in that: step S4 specifically includes the following steps:
step S41: forward propagation: a multilayer perceptron is adopted, an implicit layer is assumed to be an N layer, and the method comprises the following steps: h1＝f(W1X+B1)，Hi＝f(WiHi-1+Bi),i＝2,3,…,N，O＝g(WoHN+Bo) (ii) a Wherein, W is a weight; b is an offset value; x is multilayer perceptionInputting the machine; hiI is 1,2, …, N is the output of the ith hidden layer; o is the output of the multilayer perceptron; f (-) is the activation function of the hidden layer, expressed as
g (-) is the activation function of the output layer, the output layer will select different activation functions during training and prediction, linear function during training is selected, and positive linear function is used during prediction, the expression of which is
Step S42, back propagation: the loss function of the model is mean square error MSE, and an adaptive moment estimation algorithm Adam is used.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110331087.0A CN112949936A (en) | 2021-03-29 | 2021-03-29 | Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110331087.0A CN112949936A (en) | 2021-03-29 | 2021-03-29 | Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112949936A true CN112949936A (en) | 2021-06-11 |
Family
ID=76227040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110331087.0A Pending CN112949936A (en) | 2021-03-29 | 2021-03-29 | Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112949936A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113779861A (en) * | 2021-07-23 | 2021-12-10 | 国网河北省电力有限公司电力科学研究院 | Photovoltaic power prediction method and terminal equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017161786A1 (en) * | 2016-03-23 | 2017-09-28 | 严利容 | Optical storage system operation optimization-based hybrid energy storage ratio calculation method |
| CN109711609A (en) * | 2018-12-15 | 2019-05-03 | 福州大学 | Photovoltaic plant output power predicting method based on wavelet transformation and extreme learning machine |
| CN110766134A (en) * | 2019-09-25 | 2020-02-07 | 福州大学 | Photovoltaic power station short-term power prediction method based on cyclic neural network |
| CN112149905A (en) * | 2020-09-25 | 2020-12-29 | 福州大学 | Photovoltaic power station short-term power prediction method based on wavelet transformation and wavelet neural network |
-
2021
- 2021-03-29 CN CN202110331087.0A patent/CN112949936A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017161786A1 (en) * | 2016-03-23 | 2017-09-28 | 严利容 | Optical storage system operation optimization-based hybrid energy storage ratio calculation method |
| CN109711609A (en) * | 2018-12-15 | 2019-05-03 | 福州大学 | Photovoltaic plant output power predicting method based on wavelet transformation and extreme learning machine |
| CN110766134A (en) * | 2019-09-25 | 2020-02-07 | 福州大学 | Photovoltaic power station short-term power prediction method based on cyclic neural network |
| CN112149905A (en) * | 2020-09-25 | 2020-12-29 | 福州大学 | Photovoltaic power station short-term power prediction method based on wavelet transformation and wavelet neural network |
Non-Patent Citations (1)
| Title |
|---|
| 王骏 等: "基于深度学习的多元信息嵌入推荐模型", 《计算机应用与软件》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113779861A (en) * | 2021-07-23 | 2021-12-10 | 国网河北省电力有限公司电力科学研究院 | Photovoltaic power prediction method and terminal equipment |
| CN113779861B (en) * | 2021-07-23 | 2023-08-22 | 国网河北省电力有限公司电力科学研究院 | Photovoltaic Power Prediction Method and Terminal Equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109002915B (en) | Photovoltaic power station short-term power prediction method based on Kmeans-GRA-Elman model | |
| CN111880044B (en) | Online fault positioning method for distribution network containing distributed power supply | |
| CN112257941A (en) | Photovoltaic power station short-term power prediction method based on improved Bi-LSTM | |
| CN102478584B (en) | Wind power station wind speed prediction method based on wavelet analysis and system thereof | |
| CN113141008B (en) | Data-driven power distribution network distributed new energy consumption capability assessment method | |
| CN110334870B (en) | Photovoltaic power station short-term power prediction method based on gated cyclic unit network | |
| CN110555515A (en) | Short-term wind speed prediction method based on EEMD and LSTM | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN112149905A (en) | Photovoltaic power station short-term power prediction method based on wavelet transformation and wavelet neural network | |
| CN115796004A (en) | Photovoltaic power station ultra-short term power intelligent prediction method based on SLSTM and MLSTNet models | |
| CN113516271A (en) | Wind power cluster power day-ahead prediction method based on space-time neural network | |
| CN116307291B (en) | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition | |
| CN114399081A (en) | Photovoltaic power generation power prediction method based on weather classification | |
| CN112949895A (en) | Wind speed prediction method based on dynamic extensible neural network model | |
| Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
| CN115115125A (en) | Photovoltaic power interval probability prediction method based on deep learning fusion model | |
| CN112183877A (en) | Photovoltaic power station fault intelligent diagnosis method based on transfer learning | |
| CN115409369A (en) | Comprehensive energy system reliability evaluation method based on mechanism and data hybrid driving | |
| CN112085108A (en) | Photovoltaic power station fault diagnosis algorithm based on automatic encoder and K-means clustering | |
| CN111967660B (en) | Ultra-short-term photovoltaic prediction residual error correction method based on SVR | |
| CN112949936A (en) | Short-term photovoltaic power prediction method based on similar-day wavelet transform and multilayer perceptron | |
| CN116611702A (en) | Integrated learning photovoltaic power generation prediction method for building integrated energy management | |
| CN112734073A (en) | Photovoltaic power generation short-term prediction method based on long and short-term memory network | |
| CN116167465A (en) | Solar irradiance prediction method based on multivariate time series ensemble learning | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210611 |