CN116307291B - Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition - Google Patents
Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition Download PDFInfo
- Publication number
- CN116307291B CN116307291B CN202310571856.3A CN202310571856A CN116307291B CN 116307291 B CN116307291 B CN 116307291B CN 202310571856 A CN202310571856 A CN 202310571856A CN 116307291 B CN116307291 B CN 116307291B
- Authority
- CN
- China
- Prior art keywords
- data
- photovoltaic power
- wavelet decomposition
- power generation
- component
- 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
- 238000010248 power generation Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000004458 analytical method Methods 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000012795 verification Methods 0.000 claims abstract description 9
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 238000010606 normalization Methods 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims description 10
- 101000742844 Homo sapiens RNA-binding motif protein, Y chromosome, family 1 member A1 Proteins 0.000 claims description 8
- 102100038040 RNA-binding motif protein, Y chromosome, family 1 member A1 Human genes 0.000 claims description 8
- 230000008859 change Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 238000007619 statistical method Methods 0.000 claims description 6
- 238000010219 correlation analysis Methods 0.000 claims description 5
- 101001046999 Homo sapiens Kynurenine-oxoglutarate transaminase 3 Proteins 0.000 claims description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 238000002474 experimental method Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000000053 physical method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 101001076867 Homo sapiens RNA-binding protein 3 Proteins 0.000 description 1
- 102100025902 RNA-binding protein 3 Human genes 0.000 description 1
- 241000728173 Sarima Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
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
-
- 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
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- 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
Abstract
The invention provides a distributed photovoltaic power generation prediction method and a prediction terminal based on wavelet decomposition, which relate to the technical field of photovoltaic power generation prediction and acquire historical data of a distributed photovoltaic power station in a prediction area; classifying historical data based on the test set, the training set and the verification set, and carrying out normalization processing on the classified data; performing multi-layer wavelet decomposition on the training set data to decompose the training set data into a plurality of high-frequency signals and low-frequency signals with different scales; analyzing the correlation between each component and the meteorological data by using Pearson correlation coefficients; periodically analyzing each component, and carrying out waveform analysis statistics on each component; and constructing a DBN network model, and determining a final predicted value according to the predicted value of the DBN network model through a weighted value, namely a final result of power prediction. The invention solves the problem that the photovoltaic power generation data has space-time correlation, can effectively utilize the calculation resources in the distributed scene, and improves the calculation efficiency.
Description
Technical Field
The invention relates to the technical field of photovoltaic power generation prediction, in particular to a distributed photovoltaic power generation prediction method and a prediction terminal based on wavelet decomposition.
Background
A distributed photovoltaic power generation system refers to an entirety made up of a plurality of small photovoltaic power generation systems, which are typically distributed in different locations in a city or country, such as a roof, a wall, a floor, etc. The distributed photovoltaic power generation system has the advantages of wide distribution, high reliability, good maintainability and the like, and becomes an important form of future photovoltaic power generation systems. However, due to certain uncertainty and randomness of the distributed photovoltaic power generation system, such as factors of climate change, weather fluctuation, illumination intensity change and the like, the factors can cause larger fluctuation of photovoltaic power generation capacity, and large-scale grid connection of photovoltaic power generation can cause huge impact on the power system, so that the stability and safety of the power system are damaged. Therefore, the improvement of the prediction accuracy of the photovoltaic power generation is beneficial to the economic benefit and the operation efficiency of the photovoltaic power generation station, and the operation and the optimization of the photovoltaic power generation system have important significance.
At present, most provinces are a few concentrated photovoltaic power stations and a large number of distributed photovoltaic power stations, wherein the concentrated photovoltaic power stations generally have complete weather data, historical power data and relatively accurate prediction data, but distributed output is wide in distribution, distributed data points in a certain direct jurisdiction basically can have hundreds of distributed stations, randomness and fluctuation of the distributed stations have influence on operation of a power distribution network, great challenges are brought to dispatching operation work, and accurate grasp of power characteristics and prediction directions of distributed photovoltaic becomes a problem to be solved urgently.
Existing photovoltaic power prediction methods can be broadly divided into physical methods and statistical methods. The physical method takes weather forecast and photovoltaic system data as input, calculates irradiance and model temperature of a planar array in a typical physical method, and finally calculates power output; the statistical method is based on time series analysis, causal prediction, classification models and datasets with historical power measurements, and statistics and research of the log weather forecast, data retrieved from ground or satellite images, solar irradiance sensors and PV system data, and finally power is obtained.
At present, most researches adopt a traditional time sequence method, an artificial neural network, machine learning, deep learning and other prediction models, and the traditional time sequence methods such as ARIMA, SARIMA and the like only aim at the power generation effect in a small time range, and other factors are not considered, so that the prediction precision is not high; when the artificial neural networks such as BP neural network and RBF neural network face a large-scale photovoltaic power generation system, training time and calculation cost rise exponentially, the method is not suitable for engineering application, and when single prediction algorithms such as machine learning and deep learning face a complex linear time sequence, high-order data information cannot be realized, and prediction accuracy is difficult to improve. In this case, the combined prediction model can exert the advantages of different models, but most of the combined models at present adopt two prediction models to respectively predict, and then weight the two prediction models to obtain a final prediction value, but most of the combined models do not carry out excessive processing on high-order original data.
Disclosure of Invention
The invention provides a distributed photovoltaic power generation prediction method based on wavelet decomposition, which adopts a distributed architecture, can effectively utilize computing resources in a distributed scene and improves computing efficiency.
The method comprises the following steps:
step 1: acquiring historical data of a distributed photovoltaic power station in a preset area through a sensor and a weather station;
step 2: preprocessing the acquired historical data; classifying historical data based on the test set, the training set and the verification set, and carrying out normalization processing on the classified data; performing multi-layer wavelet decomposition on the training set data to decompose the training set data into a plurality of high-frequency signals and low-frequency signals with different scales;
step 3: analyzing the correlation between each component and the meteorological data by using Pearson correlation coefficients;
step 4: periodically analyzing each component, and counting the total number s and the positions t of the wave crests and the wave troughs of the waveform analysis of each component by using a python tool i And calculates the average period of the waveform by the following formula:
in the method, a component F1, a component F2, a component F3 and a component F4 are configured, wherein the component F1 represents the power fluctuation variation of each period in one day with 3.7 hours as a period;
component F2 represents the amount of power fluctuation variation in the period of day;
component F3 represents the amount of power fluctuation change around 8 days;
the component F4 represents the overall variation trend of the power, and the correlation between the component F4 and irradiance, humidity and air temperature is obtained based on correlation analysis;
step 5: and constructing a DBN network model, constructing a network structure of the DBN network model according to different input values, and determining a final predicted value according to the predicted value of the DBN network model through a weighted value, namely a final result of power prediction.
It should be further noted that, the preprocessing mode in step 2 includes visual analysis of power data and meteorological data;
and further comprises linear distribution analysis, outlier analysis, missing value analysis and statistical analysis on the historical data.
In step 2, the abnormal data in the history data after the preprocessing is removed, or the missing data is complemented.
It should be further noted that, in the step 2, wavelet decomposition adopts the following equation:
in the middle ofφ(t) Is a scale function;ψ(t) Is a wavelet function;h(n) Andg(n) The wavelet decomposition coefficients, respectively.
It should be further noted that, step 3: the calculation formula for analyzing the correlation between each component and the meteorological data by using the Pearson correlation coefficient is as follows:
wherein X and Y are a set of two variables; k is the number of sample points; r is the ratio of the covariance to the standard deviation product between the two variables.
R in the step 3 is in a value range from-1 to +1.
It should be further noted that the DBN network model in step 5 is constructed by a plurality of RBM bottom layers and 1 BP top layer.
It should be further noted that the DBN network model network learning process is divided into two phases:
the first stage is to train each RBM from bottom to top, build a display layer v1 to train RBM1, and then train RBM2 by taking the hidden layer h1 as an input layer v2 of a lower RBM 2;
and the like, until the training of all RBMs is completed, the initial value determination of the parameters is completed;
and in the second stage, the monitoring training is carried out on the network by combining the label data, and initial value parameters of the network are adjusted from top to bottom.
It should be further noted that the method further includes a verification and adjustment mode;
the evaluation was performed using a symmetric mean absolute percentage error MAPE and a root mean square error RMSE, with the following formula:
wherein n is the number of predicted points;is the true value of the ith point; />Is the i-th point predicted value.
The invention also provides a prediction terminal, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the steps of the distributed photovoltaic power generation prediction method based on wavelet decomposition when executing the program.
From the above technical scheme, the invention has the following advantages:
the distributed photovoltaic power generation prediction method based on wavelet decomposition provided by the invention utilizes a wavelet decomposition technology to decompose original photovoltaic power generation data into a plurality of scales, thereby overcoming the problem that the photovoltaic power generation data has space-time correlation, and simultaneously utilizes a DBN model to deeply learn and predict the decomposed data, thereby improving prediction accuracy.
The invention adopts a distributed architecture, can effectively utilize the computing resources in a distributed scene, and improves the computing efficiency. Finally, the method and the device can fully consider the influence of external factors, and improve the prediction precision. Therefore, the invention has higher practicability and economic benefit.
The invention is also based on a wavelet decomposition and Deep Belief Network (DBN) distributed photovoltaic power generation prediction system for predicting the output of distributed photovoltaic power generation. The invention can better capture the complex relation of photovoltaic power generation and improve the prediction precision and reliability by using wavelet decomposition to carry out frequency band decomposition on the photovoltaic power generation data and inputting signals of different frequency bands as characteristics into a deep confidence network.
The invention can also design an accurate prediction model aiming at the photovoltaic power generation stations at different geographic positions so as to improve the prediction precision and reliability. By processing the problems of uncertain factors, data loss and the like, the invention can better support the operation and management of the distributed photovoltaic power generation system.
Drawings
In order to more clearly illustrate the technical solutions of the present invention, the drawings that are needed in the description will be briefly introduced below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a distributed photovoltaic power generation prediction method based on wavelet decomposition;
fig. 2 is a schematic diagram of a model network structure composed of 3-layer RBMs.
Detailed Description
The distributed photovoltaic power generation prediction method based on wavelet decomposition provided by the invention utilizes a preprocessing technology, a normalization processing technology and a wavelet decomposition technology to deeply learn and predict the decomposed data by establishing a DBN network model, so that the prediction precision is improved. In addition, the invention adopts a distributed architecture, so that the computing resources in a distributed scene can be effectively utilized, and the computing efficiency is improved. Finally, the method and the device can fully consider the influence of external factors, and improve the prediction precision.
Of course, the distributed photovoltaic power generation prediction method based on wavelet decomposition of the invention can also relate to machine learning and deep learning, and generally comprises the technologies of artificial neural network, confidence network, reinforcement learning, migration learning, induction learning, teaching learning and the like.
A flowchart of a preferred embodiment of the distributed photovoltaic power generation prediction method based on wavelet decomposition of the present invention is shown in fig. 1. The distributed photovoltaic power generation prediction method based on wavelet decomposition is applied to one or more prediction terminals, wherein the prediction terminals are equipment capable of automatically performing numerical calculation and/or information processing according to preset or stored instructions, and the hardware comprises, but is not limited to, a microprocessor, an Application-specific integrated circuit (SpecificIntegratedCircuit, ASIC), a programmable gate array (Field-ProgrammableGate Array, FPGA), a digital processor (DigitalSignalProcessor, DSP), an embedded equipment and the like.
The predictive terminal is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices.
The network in which the terminal is predicted to be located includes, but is not limited to, the internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VirtualPrivateNetwork, VPN), and the like.
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flowchart of a distributed photovoltaic power generation prediction method based on wavelet decomposition in an embodiment is shown, where the method includes:
s101: historical data of the distributed photovoltaic power station in the preset area are obtained through the sensor and the weather station.
The method for acquiring the historical data of the regional distributed photovoltaic power station comprises the following steps: weather data such as temperature, humidity, wind speed, wind direction, etc.
S102: preprocessing the acquired historical data; classifying historical data based on the test set, the training set and the verification set, and carrying out normalization processing on the classified data; and carrying out multi-layer wavelet decomposition on the training set data to decompose the training set data into a plurality of high-frequency signals and low-frequency signals with different scales.
In one exemplary embodiment, preprocessing is performed on the acquired historical data, including visual analysis of power data and meteorological data, and further including linear distribution analysis, outlier analysis, missing value analysis, and statistical analysis of the historical data.
The invention can reject or correct abnormal data according to analysis conditions, complement missing data and the like, so as to improve data quality. And classifying the data into a test set, a training set and a verification set, and carrying out data normalization processing. And carrying out multi-layer wavelet decomposition on the processed training set data, decomposing the training set data into a plurality of high-frequency signals and low-frequency signals with different scales, and well decomposing the features with different scales in the original sequence.
The wavelet decomposition approximation of the invention comprises a lower frequency component of the signal, the detail comprises a higher frequency component, the approximation component and the detail component are respectively represented, the sequence can be subdivided through scaling and translation, and the wavelet decomposition approximation has excellent time-frequency characteristics in the aspect of processing complex and changeable power signals.
Wavelet decomposition uses the following equation:
in the middle ofφ(t) Is a scale function;ψ(t) Is a wavelet function;h(n) Andg(n) Is a wavelet decomposition coefficient.
S103: in order to more effectively establish a prediction model by using signals with different scales, before establishing the model, carrying out correlation analysis on the decomposed signals and meteorological data such as irradiance, temperature, humidity, wind speed, wind direction and the like, simultaneously carrying out periodic analysis, arranging all components according to the days, and analyzing the correlation between all components and the meteorological data by using Pearson correlation coefficients.
Wherein the Pearson correlation coefficient is a statistical method for measuring the degree of correlation between variables, the correlation is obtained by the R value, and the calculation formula is as follows:
wherein X and Y are a set of two variables; k is the number of sample points; r is the ratio of the covariance to the standard deviation product between the two variables. R is from-1 to +1, the closer the value is to +1, the stronger the positive correlation between variables is, and similarly, the closer the value is to-1, the stronger the negative correlation between variables is.
S104: periodically analyzing each component, and counting the total number s and the positions t of the wave crests and the wave troughs of the waveform analysis of each component by using a python tool i And calculates the average period of the waveform by the following formula
Illustratively, the power sequence wavelet decomposed components of data points of a region 1440 are analyzed for weather data correlation and self periodicity as described in Table 1.
TABLE 1
As can be seen from table 1, the component F1 represents the power fluctuation variation of each period in a day with 3.7 hours as a period, and the variation regularity of the component F1 shows that the variation regularity is poor, the randomness is strong, the complexity is high, the influence of high-frequency emergency is large, and the correlation with irradiance, air temperature and humidity is not large; f2 periodicity is obvious compared with F1, and power fluctuation changes taking a day as a period; component F3 represents the amount of power fluctuation change around 8 days; the component F4 represents the overall variation trend of the power, and the correlation between the component F4 and irradiance, humidity and air temperature is obtained based on correlation analysis; because the F4 period is larger, the fluctuation change is gentle, the overall change trend of the power is shown, and the correlation analysis shows that the correlation between the power and irradiance, the correlation between the humidity and the air temperature are stronger, and the irradiance correlation is highest.
Thus, by the above analysis, the input set of component modeling can be determined, F1 input is the first 4 hours power value and the predicted day type of the day before the predicted point; f2 is input as a power value of 1 day before the predicted point and a predicted day type; f3 takes eight days as the minimum unit, and all represent the power change in eight days, and has obvious periodicity, so that the input is the power value and the predicted day type at the same time of the week before the predicted point; f4 is used as a trend component and is significantly affected by irradiance, so the input selects power for the same moment in time for a plurality of consecutive days.
S105: and constructing a DBN network model, constructing a network structure of the DBN network model according to different input values, and determining a final predicted value according to the predicted value of the DBN network model through a weighted value, namely a final result of power prediction.
Specifically, the construction of the DBN network model finally determines an optimal model structure through a large number of experiments and debugging, and the optimal network structure of the above components is shown in table 2:
TABLE 2
The DBN is a network structure constructed by a plurality of RBM bottom layers and 1 BP top layer, and comprises 1 display layer and a plurality of hidden layers, wherein the more the hidden layers are, the higher the complexity is. Fig. 2 is a model network structure consisting of a 3-layer RBM. Wherein v1 and h1 are used as RBM1, RBM2 and RBM3 are stacked in turn on the basis, and the last layer is a BP layer.
The network learning process is divided into two phases: the first stage is to train each RBM from bottom to top, firstly build a display layer v1 to finish training of RBM1, and then use the obtained hidden layer h1 as an input layer v2 of a lower RBM2 to finish training of RBM 2. And so on until training of all RBMs is completed. This process is pre-training, completing the initial value determination of the parameters. And in the second stage, the monitoring training is carried out on the network by combining the label data, and initial value parameters of the network are adjusted from top to bottom, so that model errors are reduced, and the final prediction accuracy of power prediction is improved.
The invention also relates to verification and adjustment of the distributed photovoltaic power generation prediction method. In the verification and adjustment process, in order to compare the performance of network prediction, the symmetrical mean absolute percentage error MAPE and root mean square error RMSE are used for evaluation, and the formula is as follows:
wherein n is the number of predicted points;is the true value of the ith point; />Is the i-th point predicted value.
In this way, the distributed photovoltaic power generation prediction method based on wavelet decomposition processes photovoltaic power generation historical data, and can decompose the data into a plurality of high-frequency signals and low-frequency signals with different scales, so that the data can be better analyzed and modeled. The DBN model is utilized to model the data, so that the relation between the data can be captured better, and the accuracy and stability of prediction are improved.
In addition, the invention optimizes and processes the data through preprocessing, normalization, data segmentation and other technologies, thereby improving the quality and reliability of the data. The final predicted value is determined through the weighted value, so that the predicted results of different models can be reasonably combined, and the precision and reliability of prediction are improved.
The units and algorithm steps of each example described in the embodiments disclosed in the wavelet decomposition-based distributed photovoltaic power generation prediction method provided by the invention can be implemented in electronic hardware, computer software or a combination of the two, and in order to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functions in the above description. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. 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.
The wavelet decomposition-based distributed photovoltaic power generation prediction method provided by the present invention is the units and algorithm steps of each example described in connection with the embodiments disclosed herein, and can be implemented in electronic hardware, computer software, or a combination of both, and to clearly illustrate the interchangeability of hardware and software, the components and steps of each example have been generally described in terms of functionality in the foregoing description. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. 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.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (8)
1. A distributed photovoltaic power generation prediction method based on wavelet decomposition is characterized by comprising the following steps:
step 1: acquiring historical data of a distributed photovoltaic power station in a preset area through a sensor and a weather station;
step 2: preprocessing the acquired historical data; classifying historical data based on the test set, the training set and the verification set, and carrying out normalization processing on the classified data; performing multi-layer wavelet decomposition on the training set data to decompose the training set data into a plurality of high-frequency signals and low-frequency signals with different scales;
step 3: analyzing the correlation between each component and the meteorological data by using Pearson correlation coefficients;
step 4: periodically analyzing each component, and counting the total number s and the positions t of the wave crests and the wave troughs of the waveform analysis of each component by using a python tool i And calculates the average period of the waveform by the following formula:
in the method, as shown in table 1, the components after wavelet decomposition of the power sequence of 1440 data points in a certain area are subjected to meteorological data correlation analysis and self-periodicity analysis;
TABLE 1
Component F1 represents the amount of power fluctuation variation per cycle in one day with a period of 3.7 hours;
component F2 represents the power fluctuation variation in the period of days;
component F3 represents the amount of power fluctuation change with 8 days as a period;
component F4 represents the overall trend of the power;
determining an input set of component modeling by analysis;
step 5: constructing a DBN network model, constructing a network structure of the DBN network model according to different input values, and determining a final predicted value according to the predicted value of the DBN network model through a weighted value, namely a final result of power prediction;
wherein, the construction of the DBN network model determines the optimal model structure through experiments and debugging, and the optimal network structure of the components in the step 4 is shown in the table 2:
TABLE 2
The DBN is a network structure constructed by a plurality of RBM bottom layers and 1 BP top layer, and comprises 1 display layer and a plurality of hidden layers;
the DBN network model network learning process is divided into two stages:
the first stage is to train each RBM from bottom to top, build a display layer v1 to train RBM1, and then train RBM2 by taking the hidden layer h1 as an input layer v2 of a lower RBM 2;
and the like, until the training of all RBMs is completed, the initial value determination of the parameters is completed;
and in the second stage, the monitoring training is carried out on the network by combining the label data, and initial value parameters of the network are adjusted from top to bottom.
2. The wavelet decomposition-based distributed photovoltaic power generation prediction method of claim 1, wherein,
the preprocessing mode in the step 2 comprises visual analysis of power data and meteorological data;
and further comprises linear distribution analysis, outlier analysis, missing value analysis and statistical analysis on the historical data.
3. The wavelet decomposition-based distributed photovoltaic power generation prediction method according to claim 1, wherein in step 2, abnormal data existing in the history data after preprocessing is removed or missing data is complemented.
4. The wavelet decomposition-based distributed photovoltaic power generation prediction method of claim 1, wherein the wavelet decomposition in step 2 uses the following equation:
in the middle ofφ(t) Is a scale function;ψ(t) Is a wavelet function;h(n) Andg(n) The wavelet decomposition coefficients, respectively.
5. The wavelet decomposition-based distributed photovoltaic power generation prediction method of claim 1, wherein step 3: the calculation formula for analyzing the correlation between each component and the meteorological data by using the Pearson correlation coefficient is as follows:
wherein X and Y are a set of two variables; k is the number of sample points; r is the ratio of the covariance to the standard deviation product between the two variables.
6. The method for predicting distributed photovoltaic power generation based on wavelet decomposition according to claim 5, wherein R in step 3 has a value ranging from-1 to +1.
7. The wavelet decomposition-based distributed photovoltaic power generation prediction method according to claim 1, wherein the method further comprises verification and adjustment modes;
the evaluation was performed using a symmetric mean absolute percentage error MAPE and a root mean square error RMSE, with the following formula:
wherein n is the number of predicted points;is the true value of the ith point; />Is the i-th point predicted value.
8. A prediction terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the wavelet decomposition based distributed photovoltaic power generation prediction method according to any of claims 1 to 7 when executing the program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310571856.3A CN116307291B (en) | 2023-05-22 | 2023-05-22 | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310571856.3A CN116307291B (en) | 2023-05-22 | 2023-05-22 | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116307291A CN116307291A (en) | 2023-06-23 |
| CN116307291B true CN116307291B (en) | 2023-08-29 |
Family
ID=86820618
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310571856.3A Active CN116307291B (en) | 2023-05-22 | 2023-05-22 | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116307291B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117060373B (en) * | 2023-06-27 | 2024-03-26 | 国网信息通信产业集团有限公司 | Active power distribution network state estimation method and device based on measurement and alignment |
| CN116776645B (en) * | 2023-08-21 | 2023-10-27 | 成都协致科技有限责任公司 | Method and system for distributing environmental air monitoring stations based on wavelet analysis |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109902874A (en) * | 2019-02-28 | 2019-06-18 | 武汉大学 | A kind of micro-capacitance sensor photovoltaic power generation short term prediction method based on deep learning |
| CN110610280A (en) * | 2018-10-31 | 2019-12-24 | 山东大学 | Short-term prediction method, model, device and system for power load |
| CN111079980A (en) * | 2019-11-22 | 2020-04-28 | 天合云能源互联网技术(杭州)有限公司 | Optical power prediction method based on self-adaptive classification strategy and hybrid optimization SVR |
| CN114004424A (en) * | 2021-12-21 | 2022-02-01 | 南方电网科学研究院有限责任公司 | Photovoltaic power prediction method, system, equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11210617B2 (en) * | 2015-10-08 | 2021-12-28 | Johnson Controls Technology Company | Building management system with electrical energy storage optimization based on benefits and costs of participating in PDBR and IBDR programs |
| EP4222563A1 (en) * | 2020-10-04 | 2023-08-09 | Strong Force Iot Portfolio 2016, LLC | Industrial digital twin systems and methods with echelons of executive, advisory and operations messaging and visualization |
-
2023
- 2023-05-22 CN CN202310571856.3A patent/CN116307291B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110610280A (en) * | 2018-10-31 | 2019-12-24 | 山东大学 | Short-term prediction method, model, device and system for power load |
| CN109902874A (en) * | 2019-02-28 | 2019-06-18 | 武汉大学 | A kind of micro-capacitance sensor photovoltaic power generation short term prediction method based on deep learning |
| CN111079980A (en) * | 2019-11-22 | 2020-04-28 | 天合云能源互联网技术(杭州)有限公司 | Optical power prediction method based on self-adaptive classification strategy and hybrid optimization SVR |
| CN114004424A (en) * | 2021-12-21 | 2022-02-01 | 南方电网科学研究院有限责任公司 | Photovoltaic power prediction method, system, equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 光伏功率预测技术;龚莺飞;鲁宗相;乔颖;王强;;电力系统自动化(第04期);146-157 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116307291A (en) | 2023-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070226B (en) | Photovoltaic power prediction method and system based on convolutional neural network and meta-learning | |
| Ghaderi et al. | Deep forecast: Deep learning-based spatio-temporal forecasting | |
| CN116307291B (en) | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition | |
| CN110909919A (en) | Photovoltaic power prediction method of depth neural network model with attention mechanism fused | |
| Berrezzek et al. | Efficient Wind Speed Forecasting Using Discrete Wavelet Transform and Artificial Neural Networks. | |
| CN111027775A (en) | Step hydropower station generating capacity prediction method based on long-term and short-term memory network | |
| CN112285807A (en) | Meteorological information prediction method and device | |
| Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
| CN111625399A (en) | Method and system for recovering metering data | |
| CN114358389A (en) | Short-term power load prediction method combining VMD decomposition and time convolution network | |
| CN114399081A (en) | Photovoltaic power generation power prediction method based on weather classification | |
| CN112508279A (en) | Regional distributed photovoltaic prediction method and system based on spatial correlation | |
| Wang et al. | Hybrid deep learning model for short-term wind speed forecasting based on time series decomposition and gated recurrent unit | |
| CN116014722A (en) | Sub-solar photovoltaic power generation prediction method and system based on seasonal decomposition and convolution network | |
| CN113837434A (en) | Solar photovoltaic power generation prediction method and device, electronic equipment and storage medium | |
| CN116307287B (en) | Prediction method, system and prediction terminal for effective period of photovoltaic power generation | |
| CN113111592A (en) | Short-term wind power prediction method based on EMD-LSTM | |
| CN111488974A (en) | Deep learning neural network-based ocean wind energy downscaling method | |
| Phan et al. | Application of a new Transformer-based model and XGBoost to improve one-day-ahead solar power forecasts | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene | |
| Chang et al. | Forecasting wind power generation by a new type of radial basis function-based neural network | |
| Bantupalli et al. | Wind Speed forecasting using empirical mode decomposition with ANN and ARIMA models | |
| CN113761023A (en) | Photovoltaic power generation short-term power prediction method based on improved generalized neural network | |
| CN114444763A (en) | Wind power prediction method based on AFSA-GNN | |
| CN113112085A (en) | New energy station power generation load prediction method based on BP neural network |
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 |