CN115271242A - Training method, prediction method and device of photovoltaic power generation power prediction model - Google Patents
Training method, prediction method and device of photovoltaic power generation power prediction model Download PDFInfo
- Publication number
- CN115271242A CN115271242A CN202210970798.7A CN202210970798A CN115271242A CN 115271242 A CN115271242 A CN 115271242A CN 202210970798 A CN202210970798 A CN 202210970798A CN 115271242 A CN115271242 A CN 115271242A
- Authority
- CN
- China
- Prior art keywords
- photovoltaic power
- trained
- data
- prediction model
- prediction
- 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
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
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
-
- 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
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- 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
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/22—The renewable source being solar energy
- H02J2300/24—The renewable source being solar energy of photovoltaic origin
Abstract
The embodiment of the invention discloses a training method, a prediction method and a prediction device of a photovoltaic power generation power prediction model. Dividing the data to be trained according to the weather type to obtain various divided data to be trained; inputting various divided data to be trained into a preset photovoltaic power integration prediction model Stacking TCN-LGBM-PM; calculating each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each kind of divided data to be trained; and training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each kind of divided data to be trained and the identification photovoltaic power corresponding to each kind of divided data to be trained to obtain the trained photovoltaic power integrated prediction model corresponding to various weather types. The problem of the photovoltaic power generation power prediction accuracy is too low is solved, and the photovoltaic power generation power prediction accuracy is improved.
Description
Technical Field
The invention relates to the field of target identification, in particular to a training method, a prediction method and a prediction device of a photovoltaic power generation power prediction model.
Background
The Chinese energy structure needs to accelerate transformation speed, and new energy power generation gradually replaces a part of thermal power generation.
The photovoltaic power generation industry in China is in a high-speed development stage, and as far as 2020, the installed capacity of grid-connected solar power generation is 253.43GW. Along with the increase of the proportion of the photovoltaic installation machine, the phenomenon of light abandonment is also caused continuously, and the photovoltaic power generation consumption is very important.
The accurate photovoltaic power prediction plays a crucial role in power system scheduling, and how to improve the prediction power accuracy is a big problem in the current photovoltaic power generation prediction.
Disclosure of Invention
The embodiment of the invention provides a training method, a prediction method, a device, equipment and a storage medium of a photovoltaic power generation power prediction model, solves the problem of too low prediction precision of photovoltaic power generation power, and improves the prediction precision of the photovoltaic power generation power.
In order to solve the technical problems, the invention comprises the following steps:
in a first aspect, a training method of a photovoltaic power generation power prediction model is provided, and the method includes:
dividing data to be trained according to weather types to obtain various divided data to be trained, wherein the various divided data to be trained comprise sunny data, cloudy data and rainy data;
inputting various divided data to be trained into a preset photovoltaic power integration prediction model (Stacking TCN-LGBM-PM);
calculating each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each kind of divided data to be trained;
and training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each kind of divided data to be trained and the identification photovoltaic power corresponding to each kind of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types, wherein one weather type corresponds to one photovoltaic power integrated prediction model.
In some implementation manners of the first aspect, calculating each type of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain a photovoltaic power corresponding to each type of divided data to be trained includes:
and calculating to obtain the photovoltaic power corresponding to each kind of divided data to be trained according to the humidity, the temperature and the solar radiation quantity in each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model.
In some implementations of the first aspect, the TCN includes a hole convolution and residual module model;
the hole convolution comprises an input layer, a hidden layer and an output layer, wherein the hole convolutionThe expansion convolution is calculated asWherein the void factor d = (1, \8230; 2) L ) K is the convolution kernel size;
the residual module model comprises an activation function Relu and a cavity convolution layer;
W (1) 、W (2) is the weight matrix corresponding to the input, b is the offset vector, S (i,j) Indicating the activation function of the ith layer of the jth block.
In some implementations of the first aspect, the LGBM looks for leaves, splits based on a Leaf-wise algorithm, and loops around this, the LGBM adds a depth limit on the Leaf-wise to prevent the over-fitting phenomenon;
wherein, the target function of the LGBM is Obj (t) = L (t) + Ω (t) + c, Ω (t) represents a regular function, reflecting the complexity of the model, t represents the sampling time, c represents an additional parameter, L (t) represents a loss function,
in some implementations of the first aspect, the Stacking integration model includes a first layer predictive model including a plurality of base learners and a second layer predictive model including a meta-learner.
In a second aspect, a photovoltaic power generation power prediction method is provided, and the prediction method includes:
acquiring weather data to be predicted, and determining the weather type of the weather data;
calculating weather data based on a photovoltaic power integrated prediction model corresponding to a weather type to obtain photovoltaic power generation power, wherein the photovoltaic power integrated prediction model is obtained based on the first aspect and training methods in some implementation manners of the first aspect.
In a third aspect, a training apparatus for a photovoltaic power generation power prediction model is provided, the apparatus including:
the device comprises a dividing module, a training module and a training module, wherein the dividing module is used for dividing data to be trained according to weather types to obtain a plurality of divided data to be trained, and the plurality of divided data to be trained comprise sunny data, cloudy data and rainy data;
the input module is used for inputting various divided data to be trained into a preset photovoltaic power integration prediction model (TCN-LGBM-PM);
the calculation module is used for calculating each type of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain photovoltaic power corresponding to each type of divided data to be trained;
and the training module is used for training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each type of divided data to be trained and the identification photovoltaic power corresponding to each type of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types, wherein one weather type corresponds to one photovoltaic power integrated prediction model.
In a fourth aspect, a photovoltaic power generation power prediction apparatus is provided, the prediction apparatus including:
the determining module is used for acquiring weather data to be predicted and determining the weather type of the weather data;
the calculation module is used for calculating weather data based on a photovoltaic power integration prediction model corresponding to a weather type to obtain photovoltaic power generation power, wherein the photovoltaic power integration prediction model is obtained based on the first aspect and training methods in some implementation manners of the first aspect.
In a fifth aspect, an electronic device is provided, the device comprising: a processor and a memory storing computer program instructions;
the processor, when executing the computer program instructions, implements the first aspect and the training method in some implementations of the first aspect or implements the prediction method of the second aspect.
A sixth aspect provides a computer storage medium having stored thereon computer program instructions to be executed by a processor to implement the first aspect, and the training method in some implementations of the first aspect, or to implement the prediction method of the second aspect.
The embodiment of the invention provides a training method, a prediction method, a device, equipment and a storage medium of a photovoltaic power generation power prediction model. The method for training the photovoltaic power generation power prediction model comprises the following steps: dividing the data to be trained according to the weather type to obtain a plurality of kinds of divided data to be trained, wherein the plurality of kinds of divided data to be trained comprise sunny data, cloudy data and rainy data; inputting various divided data to be trained into a preset photovoltaic power integration prediction model (Stacking TCN-LGBM-PM); calculating each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each kind of divided data to be trained; and training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each type of divided data to be trained and the identification photovoltaic power corresponding to each type of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types for photovoltaic power prediction, wherein one weather type corresponds to one photovoltaic power integrated prediction model. The model used in the training and predicting process is integrated with the Stacking TCN-LGBM-PM integrated predicting model, the problem that the photovoltaic power generation power prediction precision is too low is solved, and the photovoltaic power generation power prediction precision is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments of the present invention will be briefly described below, and for those skilled in the art, other drawings may be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a training method of a photovoltaic power generation power prediction model according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a time convolution network structure including hole convolution according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a residual linking method according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a histogram algorithm according to an embodiment of the present invention;
FIG. 5 is a schematic view of a Leaf-wise growth mode provided by an embodiment of the present invention;
FIG. 6 is a schematic diagram of a Stacking integration model provided in an embodiment of the present invention;
FIG. 7 is a schematic diagram of a cross-validation provided by an embodiment of the present invention;
FIG. 8 is a schematic diagram of photovoltaic power generation under different weather types according to an embodiment of the present invention;
FIG. 9 is a comparative photovoltaic predicted power diagram provided by an embodiment of the present invention;
FIG. 10 is a schematic diagram illustrating photovoltaic power prediction and error comparison for seven consecutive days according to an embodiment of the present invention;
FIG. 11 is a MAPE value statistical plot of prediction results of different models provided by embodiments of the present invention;
FIG. 12 is a statistical plot of RMSE values for different model prediction results provided by embodiments of the present invention;
FIG. 13 is a histogram of predicted error values for four models provided by an embodiment of the present invention;
FIG. 14 is a flow chart of a mechanical model and data-driven joint photovoltaic power prediction provided by an embodiment of the present invention;
FIG. 15 is a schematic flow chart of a photovoltaic power generation power prediction method according to an embodiment of the present invention;
fig. 16 is a schematic structural diagram of a training device of a photovoltaic power generation power prediction model according to an embodiment of the present invention;
fig. 17 is a schematic structural diagram of a photovoltaic power generation power prediction apparatus according to an embodiment of the present invention;
fig. 18 is a block diagram of a computing device provided by an embodiment of the invention.
Detailed Description
Features and exemplary embodiments of various aspects of the present invention will be described in detail below, and in order to make objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not to be construed as limiting the invention. It will be apparent to one skilled in the art that the present invention may be practiced without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the present invention.
It should be noted that, in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising 8230; \8230;" comprises 8230; "does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
The photovoltaic power generation prediction method based on data driving is simple in modeling, mature in algorithm and high in calculation speed, but is completely based on numerical calculation, highly depends on data quality and quantity, does not consider the internal mechanism of photovoltaic power generation, and is poor in reliability of prediction results. The mechanism driving method can well reflect the photovoltaic internal power generation principle and the coupling effect, can analyze the physical essence of photovoltaic power generation, has high reliability and universality, and has the defects of high model complexity, high calculation difficulty, strong model parameter time-varying property and the like. At present, a widely applied data-driven prediction method in engineering highly depends on data quality and quantity, and the cost for acquiring comprehensive and qualified data of a new energy power generation system is high. The combined mechanism model and the data-driven algorithm can effectively reduce the dependence of prediction on data, can give consideration to the accuracy, speed and reliability of prediction under the condition of non-ideal data quality and quantity, greatly reduces the cost of acquiring data in actual production, and has remarkable economic benefit. Meanwhile, the accuracy of the short-term prediction of the new energy can be effectively improved, important support is played in the aspects of scheduling plan making, electric power market trading, load management and the like, the safe and economic operation level of a power grid is further improved, and the grid-connected service quality of a power grid company for new energy power generation enterprises and users is also improved.
The mechanism and data jointly drive a prediction mode, and a domain prior knowledge model and a data-driven deep learning algorithm can be combined. An electric power knowledge model is applied, model parameters are optimized, and model adaptability is improved; the efficiency of machine learning and data mining in data driving is improved, and the machine learning generalization risk is reduced on the premise of not increasing the number of training samples. The mechanism model and the data driving model are effectively combined, organic integration of rules and experiences is achieved, the advantages of the two models can be better integrated, better comprehensive performance can be achieved by adopting fewer data samples and a more simplified mechanism model, good prediction precision and efficiency are guaranteed, and reliability of a prediction result is effectively improved.
In the prior art, an improved clear sky power model based on online updating is provided, ultra-short-term prediction is made for photovoltaic power in small fluctuation weather, and prediction accuracy in small fluctuation weather at a scale of 3-4 h is improved well. Document [2] provides a photovoltaic output prediction combination model based on a self-adaptive fuzzy time sequence, historical power data are processed by adopting a self-adaptive algorithm, then the historical power data are clustered, domain-of-discourse is defined, divided and fuzzified, finally, prediction is carried out by combining a fuzzy time sequence method, the result is defuzzified, simulation is carried out on data with the use time interval of 15min on an actual photovoltaic experimental system, and the average absolute error (MAE) value and the average absolute percentage error (EMAPE) value respectively reach 1.038MW and 13.34%. Document [3] adopts an Extra Trees Regressor method to evaluate the feature importance according to the correlation between the photovoltaic output and the external environmental factors, after cleaning and extracting the feature quantity of the data, an SVM based on an artificial fish swarm method is adopted to classify the meteorological data, the photovoltaic power generation quantity of each category is predicted, and compared with the traditional SVM model, the prediction precision is greatly improved.
In addition, in the prior art, a photovoltaic power generation prediction system based on a T-S type fuzzy neural network is further disclosed, the organic combination of a fuzzy inference system and a neural network learning system is realized, meteorological factors are introduced, and the prediction accuracy and reliability are effectively improved. And clustering the historical processing data of the photovoltaic power station into K clusters day by day to construct a scheme of improving generalized weather mapping of one or more digital labels corresponding to weather professional weather, so that the defect of low prediction accuracy of photovoltaic power generation under a non-fine condition is overcome. And classifying the weather by adopting a weather scoring mechanism so as to classify the power generation data, and solving a scheme of probability density function estimation of each type of data through a kernel density function so as to provide a distribution rule in the statistical sense of the power generation data. The photovoltaic power generation prediction system obtains the predicted generated energy by detecting the real-time weather condition, the serial number, the longitude and latitude, the placing angle and the photovoltaic material data of the photovoltaic power generation equipment and applying a mechanism model.
In order to overcome the defects of the prior art, a photovoltaic power prediction method of a multi-model fusion Stacking integrated learning mode considering a mechanism model is provided, namely a training method and a prediction method of a photovoltaic power generation power prediction model in the scheme. Data observation and training principle differences of different algorithms are considered, advantages of each model are fully exerted, a plurality of machine learning algorithms are built, a packing integrated learning photovoltaic prediction model embedded in the photovoltaic mechanism model is combined, and a base learner of the model comprises the photovoltaic mechanism model, a LightGBM algorithm and a time convolution network algorithm. Therefore, the problem that the prediction accuracy of the photovoltaic power generation power is too low is solved, and the prediction accuracy of the photovoltaic power generation power is improved.
The technical solutions provided by the embodiments of the present invention are described below with reference to the accompanying drawings.
Fig. 1 is a schematic flowchart of a method for training a photovoltaic power generation power prediction model according to an embodiment of the present invention, where an execution subject of the method may be a processor.
As shown in fig. 1, the training method of the photovoltaic power generation power prediction model may include:
s101: and dividing the data to be trained according to the weather type to obtain various divided data to be trained, wherein the various divided data to be trained comprise sunny data, cloudy data and rainy data.
S102: and inputting various divided data to be trained into a preset photovoltaic power integration prediction model Stacking TCN-LGBM-PM.
S103: and calculating each type of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each type of divided data to be trained.
S104: and training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each kind of divided data to be trained and the identification photovoltaic power corresponding to each kind of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types, wherein one weather type corresponds to one photovoltaic power integrated prediction model.
In S101, in the process of dividing the data to be trained according to weather types, the power generation power of the photovoltaic power station is influenced by the solar irradiance received on the current day, the received solar irradiance is influenced by the weather types, and the photovoltaic power generation power fluctuation is different under different weathers. The cloud C and precipitation p in NWP can therefore be selected as weather typing factors. The method comprises the steps of dividing the cloud cover into a clear-sky model and a cloudy-day model according to the average cloud cover in the daytime, and classifying the models into a full-rain model according to the rainfall duration in the daytime. The type of weather with training data may be determined with particular reference to table 1.
TABLE 1 weather typing model
In Table 1, c 1 A typing threshold, t, for both sunny and cloudy models 1 And setting a threshold value for the rainy day model. According to the short-term weather forecast national standard and the meteorological principles, c 1 =0.7,t 1 =4。
In addition, in an embodiment, S101 may specifically be based on a preset photovoltaic power integrated prediction model, and calculate, according to humidity, temperature, and solar radiation amount in each kind of divided data to be trained, photovoltaic power corresponding to each kind of divided data to be trained.
It should be noted that, the physical model of photovoltaic power generation predicts the generated power based on the solar radiation and the photoelectric conversion characteristics. The physical property model of photovoltaic power generation is shown in formula (1).
In the formula (1), P c For photovoltaic power generation, eta is equal to 0,1]For the conversion efficiency of photovoltaic panels, T 0 ,P 0 And E 0 Reference temperature (25 ℃), reference power in standard weather and reference irradiance (1000W/m 2), respectively. Gamma is the temperature coefficient of the photovoltaic system, E i For direct solar radiation, T i Is the photovoltaic cell temperature.
The photovoltaic cell temperature can be derived from equation (2) for heat transfer to the ambient environment,
in the formula (2), c TE Is a constant factor related to the adsorption efficiency of the photovoltaic system, c w0 And c w1 Are constant heat transfer factors and convection heat factors, equal to 25W/m 2. Multidot.K and 6.84W/m 3. Multidot.s. Multidot.K, respectively. T is A Is ambient temperature, V w Is the wind speed.
It should be further noted that the main structure of the TCN model in the prediction model can be divided into a causal convolution for the sequence and a hole convolution and residual error module model for history data memory. Due to the fact that causal relationship exists between the layers of the convolution layer, more historical data can be memorized, and the method is suitable for data of the photovoltaic power station.
The structure of the time convolution network including the cavity convolution is shown in fig. 2, each TCN layer includes L convolution layers, and the expansion convolution calculation formula is:
in formula (3): void factor d = (1, \8230; 2) L ) And k is the convolution kernel size.
In addition, a schematic diagram of residual linkage is shown in fig. 3, and Dropout represents that the activation value of a certain neuron stops working with a certain probability in the neuron propagation process, so that the generalization of the model is enhanced. Relu denotes a linear rectification function used as an activation function for a neural network; DC Conv denotes a void convolution layer.
Equations (4), (5) represent the activation function of TCN:
W (1) 、W (2) is the weight matrix corresponding to the input, b is the offset vector, S (i,j) And (3) representing an activation function of the ith layer of the jth block, wherein a formula (4) represents a result of hole convolution at the time t, and a formula (5) represents a result after residual error linkage is added.
In an embodiment, the LightGBM model in the prediction model employs a histogram-based algorithm to alleviate the influence of high-dimensional data on prediction, so that the calculation speed is increased, and the phenomenon of overfitting of the prediction model is avoided. The basic idea of histograms consists in converting successive floating point eigenvalues into k integers, giving k to a "bucket" (bin), and constructing a histogram of width k. The structure is shown in fig. 4.
In addition, in the optimization process, the LGBM adopts a Leaf-wise algorithm to find a proper Leaf,
then split and cycle through this as shown in figure 5. The LGBM adds a depth limit on the Leaf-wise to prevent overfitting. The target function of LGBM is as follows:
Obj(t)=L(t)+Ω(t)+c (6)
in the formula: Ω (t) represents a canonical function, reflecting the complexity of the model. t represents a sampling time. c represents an additional parameter, avoiding overfitting and optimizing the tree depth.
L (t) represents a loss function by describing the actual values y of the N sampling points i And predicted valuesTo reflect the fitness of the model. The definition is as follows:
residual information of previous learners is transmitted by serially coupling the regression trees. Final outputGenerated by the accumulation of the remaining trees.
Furthermore, in the present disclosure, a gradient-based one-sided sampling algorithm (GOSS) is employed. During sampling, the GOSS can completely reserve the large-gradient samples meeting the conditions as sampled data, and adopts a random sampling mode for the samples with smaller gradients, so that the data with insufficient training can be kept to pay more attention in the next training, and the sample distribution can not be greatly changed.
A mutually exclusive feature binding method (EFB) is also employed. After GOSS sampling, the dimensionality of the features is reduced by binding mutually exclusive features to prevent dimensionality disasters and improve the calculation efficiency. Because most of high-dimensional data is sparse data, most of the features in the feature space are mutually exclusive, the mutually exclusive features can be bound together to form a new feature to reduce the feature dimension.
In an embodiment, in the Stacking integrated learning framework in the photovoltaic power integrated prediction model, the Stacking integrated model divides an original data set into a plurality of sub data sets, and inputs the sub data sets to each base learner of the first-layer prediction model, each base learner predicts and outputs a respective result to form a new data set, inputs the new data set to the second-layer prediction model for training, predicts and outputs a final result, and the structure is shown in fig. 6. The Stacking integration model generalizes output results output by a plurality of models, learns combined information among features, and effectively improves the overall prediction precision.
In one embodiment, the present disclosure employs a cross-validation approach to evaluate the predictive performance of a model. The data set is divided into a plurality of subsets through cross validation, evaluation results are fused, variance of model prediction results is reduced, generalization capability of the model is improved, and over-fitting is avoided. The process is shown in fig. 7, taking 3-fold as an example.
In order to prove the accuracy of the photovoltaic power integration prediction model obtained by the training method, in one embodiment, the method also performs policy analysis, in a simulation experiment, 2019 photovoltaic and meteorological data of a photovoltaic system in a certain province are selected to verify the performance of the model, the sampling time is 6 to 18 minutes per day.
The photovoltaic output power is closely related to the meteorological conditions (irradiance, humidity, temperature, etc.). And selecting photovoltaic power generation capacity under three typical weather conditions in 4 months for comparison. Fig. 8 (a) and (b) show photovoltaic power generation for three weather types.
Fig. 8 shows that the output power of a photovoltaic power station is proportional to the irradiance of the day. And in sunny days and cloudy days, the photovoltaic output power curve is smooth, and the maximum output power can be obtained in sunny days. The photovoltaic power curve has large fluctuation in rainy days and strong nonlinearity.
FIG. 9 shows the photovoltaic predicted power comparison, cloudy (a), sunny (b) and rainy (c), and Table 2 shows the predicted error for three weather types
TABLE 2
Fig. 9 shows the actual photovoltaic output and the predicted results for each weather. Under sunny days and cloudy weather conditions, the photovoltaic power generation capacity is positively correlated with solar radiation. And under the condition of overcast and rainy weather, the correlation between photovoltaic power generation and solar radiation is weaker. In order to verify the prediction performance of the Stacking model fusion, a data-driven algorithm and a mechanism model are selected and compared. Under sunny days and cloudy days, the prediction results of all models are close to the actual values. Under the condition of overcast and rainy weather, the photovoltaic power fluctuation is large, and especially when the photovoltaic power suddenly changes, the prediction error of the single mechanism model is large. This is because some weather factors cannot be taken into account by the mechanism model, resulting in low modeling accuracy. The proposed model takes into account the internal mechanisms of the PV, which results in a slight improvement of its predictive performance.
Table 2 summarizes the numerical results of RMSE and MAPE for each method. The proposed model may provide better prediction results than mechanistic models and data-driven models under different weather types. By comparing the forecast results in different weather, the forecast effect in sunny days is better, and the forecast effect in rainy days is generally poorer. For RMSE, the error of this method was 3.50% (sunny day), 6.45%
(cloudy) and 16.41% (rainy).
The photovoltaic power prediction and error pair ratios for seven consecutive days are shown in fig. 10, the MAPE values of the model predictions are shown in fig. 11, and the RMSE values of the model predictions are shown in fig. 12.
Table 3 shows the prediction error for seven consecutive days, summarizing the numerical results for MAPE and RMSE for each method. The proposed method may provide better prediction results than a single model under different weather conditions. The results indicated a MAPE value of 8.91% for the entire predicted days of the method. LGBM (17.07%) showed an increase in TCN (20.66%) and PM model (73.39%) of 47.8%, 56.9% and 87.9%, respectively. Due to the fact that the physical model deviation point is large under the rainy and cloudy weather conditions, the overall prediction precision is reduced. In addition, the proposed method has an RMSE value of 25.03MW, which is superior to other single data-driven and mechanistic models.
TABLE 3
The error indicator boxplots for the four models are shown in fig. 13. As can be seen from fig. 13, the photovoltaic power integration prediction model trained in the present disclosure has the smallest error range in most cases. For TCN and LGBM, the mean and error ranges are smaller than for the mechanism model. The data and mechanism jointly drive the model to reduce the error range, thereby weakening the randomness of the PM model. The greatest randomness comes from the mechanistic model, due to the degree of modeling refinement. It can also be seen that the distribution interval of the mean to the lower edge of the box plot of the proposed model is the closest, which indicates that there are more predicted point errors close to 0. Therefore, the integrated prediction model of the Stacking TCN-LGBM-PM is fused in the method, the problem that the prediction accuracy of the photovoltaic power generation power is too low is solved, and the prediction accuracy of the photovoltaic power generation power is improved.
In one embodiment, the Stacking multi-model fusion short-term photovoltaic power generation prediction considering the mechanism model is introduced based on the principle, a photovoltaic physical power generation model is embedded into a layer of prediction model of a Stacking integrated model and is set as one of base learners, and the Stacking multi-model fusion photovoltaic power generation prediction model considering the mechanism model is constructed. The implementation process of short-term photovoltaic power generation power prediction is given below, and the flow chart is shown in fig. 14.
The method comprises the following steps:
(1) And dividing the weather into three weather models of sunny days, cloudy days and rainy days based on the NWP value.
(2) And judging the weather type of the day to be detected, and selecting the historical photovoltaic power data and the historical NWP value of the nearest similar day according to the weather type of the day to be detected.
(3) And inputting the classified data set into each base learner in a layer of prediction model of the Stacking integration model.
(4) And obtaining the prediction result of each base learner, constructing a new training set and inputting the new training set into the two-layer prediction model.
(5) And obtaining a prediction result of the two-layer element learner, namely the final photovoltaic power generation power prediction value.
Fig. 15 is a schematic flow chart of a photovoltaic power generation power prediction method provided in an embodiment of the present invention, and as shown in fig. 15, the prediction method includes:
s201: and acquiring weather data to be predicted, and determining the weather type of the weather data.
S202: calculating weather data based on a photovoltaic power integrated prediction model corresponding to the weather type to obtain photovoltaic power generation power, wherein the photovoltaic power integrated prediction model is obtained based on any one of the training methods in fig. 1.
The photovoltaic power is predicted based on the integrated prediction model fused with the Stacking TCN-LGBM-PM, the problem that the prediction accuracy of the photovoltaic power generation power is too low can be solved, and the prediction accuracy of the photovoltaic power generation power is improved.
Corresponding to the flow schematic diagram of the training method of the photovoltaic power generation power prediction model shown in fig. 1, the embodiment of the invention also provides a training device of the photovoltaic power generation power prediction model. Fig. 16 is a schematic structural diagram of a training apparatus for a photovoltaic power generation power prediction model according to an embodiment of the present invention, and as shown in fig. 16, the training apparatus for a photovoltaic power generation power prediction model may include:
the dividing module 301 is configured to divide the data to be trained according to the weather type to obtain a plurality of divided data to be trained, where the plurality of divided data to be trained include sunny data, cloudy data, and rainy data.
The input module 302 is configured to input the multiple types of divided data to be trained into a preset photovoltaic power integration prediction model Stacking TCN-LGBM-PM.
The calculating module 303 is configured to calculate each type of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain a photovoltaic power corresponding to each type of divided data to be trained.
The training module 304 is configured to train the photovoltaic power integration prediction model according to the calculated photovoltaic power corresponding to each type of divided data to be trained and the calculated identification photovoltaic power corresponding to each type of divided data to be trained, so as to obtain the trained photovoltaic power integration prediction models corresponding to multiple weather types, where one weather type corresponds to one photovoltaic power integration prediction model.
In the training process of the training device of the photovoltaic power generation power prediction model, the Stacking TCN-LGBM-PM integrated prediction model is fused, the problem that the photovoltaic power generation power prediction precision is too low is solved, and the photovoltaic power generation power prediction precision is improved.
Corresponding to the flow schematic diagram of the photovoltaic power generation power prediction method shown in fig. 15, an embodiment of the present invention further provides a photovoltaic power generation power prediction apparatus. Fig. 17 is a schematic structural diagram of a photovoltaic power generation power prediction apparatus provided in an embodiment of the present invention, and as shown in fig. 17, the photovoltaic power generation power prediction apparatus may include:
the determining module 401 is configured to obtain weather data to be predicted, and determine a weather type of the weather data.
A calculating module 402, configured to calculate weather data based on a photovoltaic power integration prediction model corresponding to a weather type to obtain photovoltaic power generation power, where the photovoltaic power integration prediction model is obtained based on any one of the training methods in fig. 1.
The photovoltaic power is predicted based on the integrated prediction model fused with the Stacking TCN-LGBM-PM, the problem that the prediction accuracy of the photovoltaic power generation power is too low can be solved, and the prediction accuracy of the photovoltaic power generation power is improved.
Fig. 18 is a block diagram of a hardware architecture of a computing device according to an embodiment of the present invention. As shown in fig. 18, computing device 700 includes an input device 701, an input interface 702, a central processor 703, a memory 704, an output interface 705, and an output device 707. The input interface 702, the central processing unit 703, the memory 704, and the output interface 705 are connected to each other via a bus 710, and the input device 701 and the output device 706 are connected to the bus 710 via the input interface 702 and the output interface 705, respectively, and further connected to other components of the computing device 700.
The computing device shown in fig. 18 may also be implemented as a training device of a photovoltaic generation power prediction model, or alternatively, a photovoltaic generation power prediction device, and the computing device may include: a processor and a memory storing computer executable instructions; when the processor executes the computer-executable instructions, the training method of the photovoltaic power generation power prediction model provided by the embodiment of the invention can be realized, or the photovoltaic power generation power prediction method provided by the embodiment of the invention can be realized.
An embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium has computer program instructions stored thereon; when executed by a processor, the computer program instructions implement the training method of the photovoltaic power generation power prediction model provided by the embodiment of the present invention, or implement the photovoltaic power generation power prediction method provided by the embodiment of the present invention.
It is to be understood that the invention is not limited to the precise arrangements and instrumentalities shown. A detailed description of known methods is omitted herein for the sake of brevity. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present invention are not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications and additions or change the order between the steps after comprehending the spirit of the present invention.
The functional blocks shown in the above-described structural block diagrams may be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it may be, for example, an electronic Circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, plug-in, function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted by a data signal carried in a carrier wave over a transmission medium or a communication link. A "machine-readable medium" may include any medium that can store or transfer information. Examples of a machine-readable medium include electronic circuits, semiconductor Memory devices, read-Only memories (ROMs), flash memories, erasable Read-Only memories (EROMs), floppy disks, compact disk Read-Only memories (CD-ROMs), optical disks, hard disks, optical fiber media, radio Frequency (RF) links, and so forth. The code segments may be downloaded via computer networks such as the internet, intranet, etc.
It should also be noted that the exemplary embodiments noted in this patent describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above-described steps, that is, the steps may be performed in the order mentioned in the embodiments, may be performed in an order different from the order in the embodiments, or may be performed simultaneously.
Aspects of the present disclosure are described above with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, implement the functions/acts specified in the flowchart and/or block diagram block or blocks. Such a processor may be, but is not limited to, a general purpose processor, a special purpose processor, an application specific processor, or a field programmable logic circuit. It will also be understood that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware for performing the specified functions or acts, or combinations of special purpose hardware and computer instructions.
As described above, only the specific embodiments of the present invention are provided, and it can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system, the module and the unit described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. It should be understood that the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive various equivalent modifications or substitutions within the technical scope of the present invention, and these modifications or substitutions should be covered within the scope of the present invention.
Claims (10)
1. A training method of a photovoltaic power generation power prediction model is characterized by comprising the following steps:
dividing data to be trained according to weather types to obtain various divided data to be trained, wherein the various divided data to be trained comprise sunny data, cloudy data and rainy data;
inputting the various divided data to be trained into a preset photovoltaic power integration prediction model Stacking TCN-LGBM-PM;
calculating each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each kind of divided data to be trained;
and training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each kind of divided data to be trained and the identification photovoltaic power corresponding to each kind of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types, wherein one weather type corresponds to one photovoltaic power integrated prediction model.
2. The method according to claim 1, wherein the step of calculating each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain the photovoltaic power corresponding to each kind of divided data to be trained comprises:
and calculating to obtain the photovoltaic power corresponding to each kind of divided data to be trained according to the humidity, the temperature and the solar radiation quantity in each kind of divided data to be trained based on a preset photovoltaic power integrated prediction model.
3. The method of claim 1 or 2, wherein the TCN comprises a hole convolution and residual module model;
the hole convolution comprises an input layer, a hidden layer and an output layer, and the calculation formula of the expansion convolution in the hole convolution isWherein the void factor d = (1, \8230; 2) L ) K is the convolution kernel size;
the residual module model comprises an activation function Relu and a hole convolution layer;
4. The method of claim 1, wherein the LGBM is based on a Leaf-wise algorithm to find leaves, splits, and loops accordingly, the LGBM adds a depth limit on the Leaf-wise to prevent overfitting;
5. the method of claim 1, wherein the Stacking integration model comprises a first layer prediction model and a second layer prediction model, wherein the first layer prediction model comprises a plurality of base learners and the second layer prediction model comprises a meta-learner.
6. A photovoltaic power generation power prediction method is characterized by comprising the following steps:
acquiring weather data to be predicted, and determining the weather type of the weather data;
calculating the weather data based on a photovoltaic power integration prediction model corresponding to the weather type to obtain photovoltaic power generation power, wherein the photovoltaic power integration prediction model is obtained based on the training method of any one of claims 1 to 5.
7. A training device of a photovoltaic power generation power prediction model is characterized by comprising:
the device comprises a dividing module, a training module and a training module, wherein the dividing module is used for dividing data to be trained according to weather types to obtain a plurality of divided data to be trained, and the plurality of divided data to be trained comprise sunny data, cloudy data and rainy data;
the input module is used for inputting the various divided data to be trained into a preset photovoltaic power integration prediction model (Stacking TCN-LGBM-PM);
the calculating module is used for calculating each type of divided data to be trained based on a preset photovoltaic power integrated prediction model to obtain photovoltaic power corresponding to each type of divided data to be trained;
and the training module is used for training the photovoltaic power integrated prediction model according to the calculated photovoltaic power corresponding to each type of divided data to be trained and the identification photovoltaic power corresponding to each type of divided data to be trained to obtain the trained photovoltaic power integrated prediction models corresponding to multiple weather types, wherein one weather type corresponds to one photovoltaic power integrated prediction model.
8. A photovoltaic power generation power prediction apparatus, characterized in that the prediction apparatus comprises:
the determining module is used for acquiring weather data to be predicted and determining the weather type of the weather data;
a calculation module, configured to calculate the weather data based on a photovoltaic power integration prediction model corresponding to the weather type to obtain photovoltaic power generation power, where the photovoltaic power integration prediction model is obtained based on the training method according to any one of claims 1 to 5.
9. An electronic device, characterized in that the electronic device comprises: a processor and a memory storing computer program instructions;
the processor, when executing the computer program instructions, implements a method of training a photovoltaic power generation power prediction model according to any one of claims 1-5, or implements a method of photovoltaic power generation power prediction according to claim 6.
10. A computer storage medium having stored thereon computer program instructions which, when executed by a processor, implement a method of training a photovoltaic power generation prediction model according to any one of claims 1-5, or implement a method of photovoltaic power generation prediction according to claim 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210970798.7A CN115271242A (en) | 2022-08-13 | 2022-08-13 | Training method, prediction method and device of photovoltaic power generation power prediction model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210970798.7A CN115271242A (en) | 2022-08-13 | 2022-08-13 | Training method, prediction method and device of photovoltaic power generation power prediction model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115271242A true CN115271242A (en) | 2022-11-01 |
Family
ID=83751583
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210970798.7A Pending CN115271242A (en) | 2022-08-13 | 2022-08-13 | Training method, prediction method and device of photovoltaic power generation power prediction model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115271242A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116050187A (en) * | 2023-03-30 | 2023-05-02 | 国网安徽省电力有限公司电力科学研究院 | TS fuzzy outlier self-correction method and system for second-level photovoltaic power prediction |
-
2022
- 2022-08-13 CN CN202210970798.7A patent/CN115271242A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116050187A (en) * | 2023-03-30 | 2023-05-02 | 国网安徽省电力有限公司电力科学研究院 | TS fuzzy outlier self-correction method and system for second-level photovoltaic power prediction |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tang et al. | Short‐term power load forecasting based on multi‐layer bidirectional recurrent neural network | |
CN113962364B (en) | Multi-factor power load prediction method based on deep learning | |
Wang et al. | Adaptive learning hybrid model for solar intensity forecasting | |
Wu et al. | Evolving RBF neural networks for rainfall prediction using hybrid particle swarm optimization and genetic algorithm | |
CN112116144B (en) | Regional power distribution network short-term load prediction method | |
CN112215442B (en) | Method, system, device and medium for predicting short-term load of power system | |
Liu et al. | Photovoltaic generation power prediction research based on high quality context ontology and gated recurrent neural network | |
CN114792156A (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
Niu et al. | Short-term wind speed hybrid forecasting model based on bias correcting study and its application | |
CN115860177A (en) | Photovoltaic power generation power prediction method based on combined machine learning model and application thereof | |
Li et al. | Short term prediction of photovoltaic power based on FCM and CG-DBN combination | |
CN115115125A (en) | Photovoltaic power interval probability prediction method based on deep learning fusion model | |
Bai et al. | Improved Markov‐chain‐based ultra‐short‐term PV forecasting method for enhancing power system resilience | |
CN115186923A (en) | Photovoltaic power generation power prediction method and device and electronic equipment | |
CN115271242A (en) | Training method, prediction method and device of photovoltaic power generation power prediction model | |
Goh et al. | Hybrid SDS and WPT-IBBO-DNM Based Model for Ultra-short Term Photovoltaic Prediction | |
CN114611757A (en) | Electric power system short-term load prediction method based on genetic algorithm and improved depth residual error network | |
Cui et al. | A VMD-MSMA-LSTM-ARIMA model for precipitation prediction | |
CN111697560B (en) | Method and system for predicting load of power system based on LSTM | |
CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
CN114492945A (en) | Short-term photovoltaic power prediction method, medium and equipment in electric power market background | |
CN116502074A (en) | Model fusion-based photovoltaic power generation power prediction method and system | |
Yang | Short-term load monitoring of a power system based on neural network | |
Yuan et al. | Short-term wind speed forecasting using STLSSVM hybrid model | |
CN113723670B (en) | Photovoltaic power generation power short-term prediction method with variable time window |
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 |