CN115952921A

CN115952921A - Photovoltaic energy power prediction method and device, electronic equipment and storage medium

Info

Publication number: CN115952921A
Application number: CN202310089213.5A
Authority: CN
Inventors: 陈凤超; 邹钟璐; 赵瑞锋; 邱泽坚; 段孟雍; 钟志明; 何毅鹏; 饶欢; 邓景柱; 刘铮
Original assignee: Guangdong Power Grid Co Ltd; Dongguan Power Supply Bureau of Guangdong Power Grid Co Ltd
Current assignee: Guangdong Power Grid Co Ltd; Dongguan Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority date: 2023-02-06
Filing date: 2023-02-06
Publication date: 2023-04-11

Abstract

The invention discloses a photovoltaic energy power prediction method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring historical data of operation of each distributed solar power station; wherein the historical data consists of power data and meteorological data; combining every two data in the historical data to obtain target data; processing the historical data and the target data to obtain training sample characteristics; and adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction. According to the technical scheme, the original features can be combined into new features, and then the model is trained, so that high repetition rate feature training becomes possible, and meanwhile, the accuracy of photovoltaic energy power prediction can be improved.

Description

Photovoltaic energy power prediction method and device, electronic equipment and storage medium

Technical Field

The invention relates to the technical field of energy prediction, in particular to a photovoltaic energy power prediction method, a photovoltaic energy power prediction device, electronic equipment and a storage medium.

Background

With the increase of energy demand and the increasingly prominent environmental problems, the proportion of renewable energy sources in a plurality of energy sources is increasingly improved, and the energy proportion of solar energy sources in a power grid power generation device is also improved due to the characteristics of wide distribution range, high energy utilization rate, cleanness, no pollution and the like. However, the photovoltaic power generation device has strong randomness and volatility due to weather, so that the real-time scheduling of the power grid can be challenged when a large number of photovoltaic power generation devices are connected to the grid. Therefore, the power prediction level of the solar power generation device is improved, and the method has important significance.

The existing prediction methods can be divided into physical prediction and statistical prediction models according to prediction models. The physical prediction model method is used for collecting current electric quantity information and meteorological condition information in the photovoltaic power station, and estimating future output power by establishing a high-order function and fitting a power generation power curve in combination with historical power conditions.

For a prediction method based on a physical prediction model, the prior art mainly faces two problems of low precision and great operation difficulty: the physical prediction model prediction mode is based on the current power station physical information and weather forecast information, although the calculated amount is small, the model parameters need to be manually adjusted according to the physical information of different power stations in operation, the operation difficulty is high, and the accuracy of the parameters and the weather forecast directly influences the accuracy of prediction.

Disclosure of Invention

The invention provides a photovoltaic energy power prediction method, a photovoltaic energy power prediction device, electronic equipment and a storage medium, which can combine original features into new features, further train a model, enable high repetition rate feature training to be possible, and simultaneously improve the accuracy of photovoltaic energy power prediction.

According to an aspect of the invention, a photovoltaic energy power prediction method is provided, the method comprising:

acquiring historical data of operation of each distributed solar power station; wherein the historical data consists of power data and meteorological data;

combining every two data in the historical data to obtain target data;

processing the historical data and the target data to obtain training sample characteristics;

and adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction.

According to another aspect of the present invention, there is provided a photovoltaic energy power prediction apparatus, comprising:

the historical data acquisition module is used for acquiring the running historical data of each distributed solar power station; wherein the historical data consists of power data and meteorological data;

the data combination module is used for combining each data in the historical data in pairs to obtain target data;

a training sample characteristic obtaining module, configured to process the historical data and the target data to obtain training sample characteristics;

and the photovoltaic energy power prediction module is used for adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction.

According to another aspect of the present invention, there is provided an electronic apparatus including:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,

the memory stores a computer program executable by the at least one processor, the computer program being executable by the at least one processor to enable the at least one processor to perform a method of predicting photovoltaic energy power according to any embodiment of the invention.

According to another aspect of the present invention, there is provided a computer readable storage medium storing computer instructions for causing a processor to implement a method for photovoltaic energy power prediction according to any one of the embodiments of the present invention when executed.

According to the technical scheme, historical data of operation of each distributed solar power station is obtained, then each data in the historical data is combined in pairs to obtain target data, the historical data and the target data are processed to obtain training sample characteristics, the photovoltaic energy power prediction model is adjusted according to the training sample characteristics, and the updated photovoltaic energy power prediction model is obtained and used for photovoltaic energy power prediction. According to the technical scheme, the original features can be combined into new features, and then the model is trained, so that high repetition rate feature training becomes possible, and meanwhile, the accuracy of photovoltaic energy power prediction can be improved.

It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present invention, nor do they necessarily limit the scope of the invention. Other features of the present invention will become apparent from the following description.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 is a flowchart of a photovoltaic energy power prediction method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a data flow path of a dual-platform prediction system according to an embodiment of the present disclosure;

fig. 3 is a flowchart illustrating a work of a cloud computing center according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of a photovoltaic power prediction process according to a second embodiment of the present invention;

fig. 5 is a flowchart of a photovoltaic energy power prediction model training process according to a third embodiment of the present invention;

FIG. 6 is a flowchart of a training process of a back propagation neural network provided in the third embodiment of the present application;

fig. 7 is a schematic structural diagram of a photovoltaic energy power prediction apparatus according to a fourth embodiment of the present invention;

fig. 8 is a schematic structural diagram of an electronic device implementing the photovoltaic energy power prediction method according to the embodiment of the present invention.

Detailed Description

In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It is to be understood that the terms "target" and the like in the description and claims of the present invention and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in other sequences than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Example one

Fig. 1 is a flowchart of a photovoltaic energy power prediction method according to an embodiment of the present invention, where the embodiment is applicable to a situation of predicting photovoltaic energy power, and the method may be executed by a photovoltaic energy power prediction apparatus, where the photovoltaic energy power prediction apparatus may be implemented in a form of hardware and/or software, and the photovoltaic energy power prediction apparatus may be configured in an electronic device.

As shown in fig. 1, the method includes:

s110, acquiring historical data of operation of each distributed solar power station; wherein the historical data consists of power data and meteorological data.

In the scheme, the solar energy source has the characteristics of wide distribution range, high energy utilization rate, cleanness, no pollution and the like. Photovoltaic energy can be collected based on a solar power station for load use or stored in a storage battery for later use.

In this embodiment, the historical data of the operation of the solar power plant may include conversion efficiency, voltage, field temperature, plate temperature, humidity, air pressure, power, current, wind speed, average power, wind direction, sunlight intensity, sunlight angle, altitude, and the difference between the latitude and longitude of the current solar power plant and each distributed solar power plant. The conversion efficiency may be a result obtained by dividing the output electric power of the solar cell panel by the input optical power. The plate temperature refers to the temperature measured by the temperature sensor bonded on the solar panel assembly back plate.

The historical data of the operation of the solar power station can be collected based on various sensors.

Specifically, fig. 2 is a schematic diagram of a data flow path of a dual-platform prediction system according to an embodiment of the present disclosure, and as shown in fig. 2, photovoltaic energy power prediction is implemented by using a plurality of edge power prediction platforms and a cloud computing center. The edge power prediction platforms are respectively established near the solar power station and can be connected with each other, namely, the edge power prediction platforms can be connected with each other through a network. The edge power prediction platform is mainly used for collecting data of operation of the solar power station, collected by the sensor, predicting the output power of the solar power station and sending the collected data to the cloud computing center. And the cloud computing center is responsible for storing the collected data, processing the data and judging the optimal characteristics and the optimal network structure required by power prediction of each distributed solar power station. The cloud computing center only needs data storage and data analysis of a fixed algorithm, is simple in functional logic, and can be implemented according to the existing cloud platform supporting storage and computing. The edge power prediction platform can require the cloud computing center to update the photovoltaic energy power prediction model on line at a certain frequency. For example, the cloud computing center may be required to update the photovoltaic energy power prediction model for one month.

In this technical solution, optionally, the obtaining of the historical data of the operation of each distributed solar power plant includes:

historical data of the operation of each distributed solar power station is obtained based on each sensor which is installed in advance.

In this scenario, as shown in fig. 2, historical data of the operation of each distributed solar power plant may be collected by various types of sensors. Wherein the meteorological data is collected by an unmanned automated meteorological station or other device built into the power plant. The specific meteorological sensor is composed of a temperature and humidity sensor, a wind speed and direction sensor, a light intensity sensor and the like. The power data may be collected by power sensors or other power devices. The specific power sensor may be constituted by a voltage sensor, a current sensor, a position sensor, or the like. The latitude and longitude difference between the current solar power station and each distributed solar power station can also be determined in advance according to the installation position of the solar power station.

By acquiring historical data of operation of the solar power station, the photovoltaic energy power prediction model can be trained based on the historical data, and accuracy of photovoltaic energy prediction can be improved.

And S120, combining every two data in the historical data to obtain target data.

In the scheme, each pair of data in the historical data is combined to obtain new data, so that the photovoltaic energy power prediction model is trained, and high-repetition-rate data training is possible. For example, voltage and humidity may be combined to obtain new data; the current can also be combined with the wind speed to obtain new data.

And S130, processing the historical data and the target data to obtain training sample characteristics.

According to the scheme, the historical data and the target data can be analyzed, and the data with large correlation with the output power of the solar power station is screened out from the historical data and the target data to serve as the training sample characteristics.

In this technical solution, optionally, the processing the historical data and the target data to obtain training sample characteristics includes steps A1 to A3:

a1, preprocessing the historical data and the target data to obtain the characteristics of a training sample to be processed;

in the scheme, historical data and target data can be processed based on a cloud computing center, and training sample characteristics are obtained. Specifically, fig. 3 is a flowchart of the work of the cloud computing center provided in the embodiment of the present application, and as shown in fig. 3, the historical data uploaded by all the edge power prediction platforms are sequentially preprocessed, and after the preprocessing, training sample features are obtained.

Specifically, a frequency distribution graph may be drawn on the historical data and the target data according to the frequency of occurrence of the numerical value in each interval, and a part of abnormal values may be deleted from the data conforming to the normal distribution or the bimodal normal distribution. Where an outlier may refer to a value outside of the standard deviation multiplied by the mean value ± 3.

Further, after deleting the abnormal value, the residual data are scaled to the range of [ -1,1] according to the range of the abnormal value, namely, the residual data are normalized to obtain the characteristics of the training sample to be processed, so that the training sample to be processed can be conveniently calculated and used subsequently.

A2, carrying out correlation analysis on the characteristics of the training sample to be processed to obtain a correlation coefficient;

the correlation analysis refers to analyzing two or more variable elements with correlation, so as to measure the degree of closeness of correlation of two factors.

Specifically, the output power may be subjected to correlation analysis respectively for the characteristics of the training samples to be processed. Specifically, the following formula is adopted to calculate the correlation coefficient:

wherein r is a correlation coefficient, X _i For the characteristics of the training sample to be processed at time i, Y _i Is the output power at the time of i,

mean values representing characteristics of a training sample to be processed>

The mean value of the output power is indicated.

And A3, screening the characteristics of the training sample to be processed according to the correlation coefficient to obtain the characteristics of the training sample.

In this scheme, after obtaining the correlation coefficients of the training sample features to be processed, the correlation coefficients may be arranged in a descending manner, and the training sample features to be processed corresponding to the correlation coefficients larger than a certain threshold are used as the training sample features. Optionally, the training sample feature to be processed corresponding to the correlation coefficient greater than 10% of the maximum correlation coefficient may be used as the training sample feature.

In this embodiment, as shown in fig. 3, after the data collected by the current edge power prediction platform is analyzed, the data collected by the next edge power prediction platform may be continuously analyzed until all the edge power prediction platforms finish analyzing.

The data are processed to obtain training sample characteristics, original characteristics can be combined into new characteristics according to relative geographic positions, and then the model is trained, so that high repetition rate characteristic training becomes possible.

S140, adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction.

Wherein training sample features may be divided into a training set and a test set. The training set is used to estimate parameters in the model, enabling the model to reflect reality and thus predict future or other unknown information, and the test set is used to evaluate the prediction performance of the model. For example, 80% of the data in the training sample features may be used as the training set and 20% of the data may be used as the test set.

In this embodiment, the photovoltaic energy power prediction model is a Back Propagation neural network, i.e., a BP (Back Propagation) neural network. The method has the characteristics of signal forward propagation and error directional propagation. The input values are passed from the input layer, through the hidden layer, and to the output layer. If the difference value between the output value and the expected value exceeds the threshold value, the error value is reversely propagated from the output layer to the hidden layer so as to adjust the network connection weight and the bias parameter and obtain the updated photovoltaic energyA power prediction model. As shown in fig. 3, the photovoltaic energy power prediction models of the edge power prediction platforms are analyzed by the computation center. Photovoltaic energy power prediction models corresponding to different edge power prediction platforms may be different. Wherein, the activation function of the network is defined as a Sigmoid function, and the formula is

The connection mode is full connection.

In the scheme, after the updated photovoltaic energy power prediction model is obtained, the output power of the solar power station can be predicted through the photovoltaic energy power prediction model corresponding to the edge power prediction platform. The data collected by the various sensors may include conversion efficiency, voltage, field temperature, plate temperature, humidity, air pressure, power, current, wind speed, average power, wind direction, sunlight intensity, sunlight angle, altitude, and the difference between the solar power station and each distributed solar power station. For a solar power station without a corresponding measuring device, only data which can be collected are collected.

Further, according to the transmission protocol of the information in each router, the edge power prediction platform uploads all data to the cloud computing center in real time and then issues the data to occupy more network resources, and actually, the data to be transmitted only need to be transmitted point to point through the distributed edge power prediction platform. And the edge power prediction platform establishes connection with the corresponding edge power prediction platform according to the characteristics required by the input nodes of the prediction model, and asks for corresponding data. Therefore, on the premise of massive photovoltaic power stations, on one hand, data are transmitted between the distributed edge power prediction platforms with a high delay requirement, and on the other hand, all data are uploaded to the cloud computing center with a low delay requirement. Therefore, the photovoltaic energy power prediction model can be executed on a small computing power platform, on one hand, high manufacturing cost is avoided, and on the other hand, the problem that all solar station prediction is invalid due to the fact that the cloud is attacked by a network is avoided.

In this embodiment, the edge power prediction platform establishes a connection with a corresponding edge power prediction platform according to the characteristics required by the input node of the prediction model, and requests for corresponding data. The corresponding relation between the edge power prediction platform and other edge power prediction platforms can be determined through correlation. For example, if the edge power prediction platform at the solar power station D obtains the output power of the solar power station D and the voltage and the illumination intensity at D through correlation calculation; and the quotient correlation between the illumination intensity at the solar power station C and the voltage at the solar power station E and the plate temperature is more than 10%, and other edge power prediction platforms corresponding to the edge power prediction platform at the position D refer to the edge power prediction platforms at the solar power stations C and E.

In this embodiment, if the solar power plant D has no device for measuring the intensity of light, the solar power plant B adjacent to the solar power plant D has. If the correlation analysis determines that the correlation coefficient between the output power of the solar power station D and the illumination intensity of the solar power station B is greater than 10%, the distributed edge power prediction platform at the solar power station B needs to transmit illumination intensity data to the distributed edge power prediction platform at the solar power station D, so that the output power prediction accuracy of the solar power station D is improved. This reduces the power plant detection equipment requirements of the power prediction algorithm, and does not need to detect all the above features for each power plant. The method is more suitable for the actual condition that the distributed power station equipment is uneven.

In this embodiment, the solar power plant power prediction system includes a plurality of edge power prediction platforms and a cloud computing center platform. And the cloud computing center provides the optimal prediction model of each edge power prediction platform through historical data. The edge power prediction platform realizes power prediction through a given model.

According to the technical scheme of the embodiment of the invention, historical data of operation of each distributed solar power station is obtained, then each data in the historical data is combined pairwise to obtain target data, the historical data and the target data are processed to obtain training sample characteristics, and the photovoltaic energy power prediction model is adjusted according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction. By executing the technical scheme, the original features can be combined into new features, and then the model is trained, so that the high repetition rate feature training is possible, and meanwhile, the accuracy of photovoltaic energy power prediction can be improved.

Example two

Fig. 4 is a schematic diagram of a photovoltaic energy power prediction process provided in the second embodiment of the present invention, and the relationship between this embodiment and the foregoing embodiments is a detailed description of a data processing process for the operation of a solar power plant. As shown in fig. 4, the method includes:

s410, acquiring historical data of operation of each distributed solar power station; wherein the historical data consists of power data and meteorological data.

S420, performing pairwise operation on each data in the historical data to obtain target data; wherein the operation comprises at least one of addition, subtraction, multiplication, and division.

In this embodiment, pairwise operation may be performed on each data in the historical data to obtain a new data. For example, the voltage and the plate temperature may be added to obtain a new data, or the voltage and the current may be divided to obtain a new data.

In this technical solution, optionally, performing pairwise operation on each data in the historical data to obtain target data, including:

and multiplying and/or dividing each data in the historical data to obtain target data.

In the scheme, each data in the historical data can be multiplied to obtain new data, and each data in the historical data can be divided to obtain new data. For example, the data acquisition system acquires data such as voltage and plate temperature for one round at 5-minute intervals. Suppose the voltage U = (U) collected during a day ₁ ,u ₂ ,K u _n ) The collected plate temperature is B = (B) ₁ ,b ₂ ,K,b _n ) The combination of voltage and plate temperature is UB = (u) ₁ *b ₁ ,u ₂ *b ₂ ,L,u _n *b _n )，U/B＝(u ₁ /b ₁ ,u ₂ /b ₂ ,L,u _n /b _n ) UB and U/B are now a new data.

The historical data are combined to obtain new data, and then the photovoltaic energy power prediction model is trained, so that high repetition rate characteristic training becomes possible.

In this technical solution, optionally, performing pairwise operation on each data in the historical data to obtain target data, further includes:

determining the order of each data in the historical data;

and multiplying and/or dividing each data according to the order of each data to obtain target data.

The order may refer to an autoregressive order, and the order of each data may be set according to a requirement. For example, the order of the voltage may be set to 2 steps, and the order of the plate temperature may be set to 1 step. The two-to-two combination order is set as the second order and the first order, and the multiplication combination of the voltage and the plate temperature is

U ² B will be treated as a new data. When the set order is three-order or one-order, the combination of the voltage and the plate temperature division is->

U ³ the/B will be a new datum. After the order is appointed, the new data is also formed by pairwise combination of historical data, wherein the pairwise combination comprises multiplication and division of the historical data and the historical data.

The historical data are combined to obtain new target data, and then the photovoltaic energy power prediction model is trained, so that high repetition rate characteristic training becomes possible.

And S430, processing the historical data and the target data to obtain training sample characteristics.

S440, adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction.

According to the technical scheme of the embodiment of the invention, the target data is obtained by acquiring the running historical data of each distributed solar power station and then performing pairwise operation on each data in the historical data, wherein the operation comprises at least one of addition, subtraction, multiplication and division. And processing the historical data and the target data to obtain training sample characteristics, and adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction. By executing the technical scheme, the original features can be combined into new features, and then the model is trained, so that the high repetition rate feature training is possible, and meanwhile, the accuracy of photovoltaic energy power prediction can be improved.

EXAMPLE III

Fig. 5 is a flowchart of a photovoltaic energy power prediction model training process provided in the third embodiment of the present invention, and a relationship between this embodiment and the foregoing embodiments is a detailed description of the photovoltaic energy power prediction model training process. As shown in fig. 5, the method includes:

s510, obtaining historical data of operation of each distributed solar power station; wherein the historical data is composed of power data and meteorological data.

S520, combining the data in the historical data in pairs to obtain target data.

S530, processing the historical data and the target data to obtain training sample characteristics.

And S540, inputting the training sample characteristics into a photovoltaic energy power prediction model to obtain output power.

In the scheme, in the back propagation neural network, the input vector is X = (X) ₁ ,x ₂ ,L,x _n ) ^T . Hidden layer output vector is Z = (Z) ₁ ,z ₂ ,L,z _r ) ^T . The output layer output vector is Y = (Y) ₁ ,y ₂ ,L,y _m ) ^T . The weight from the input layer node to the hidden layer node is w _ik . HidingThe weight from the layer node to the output layer node is v _kj . Hidden layer bias parameter is theta _k . Output layer bias parameter is gamma _j . The activation function of the hidden layer node is f ₁ . The activation function of the output layer node is f ₂ . Wherein, f ₁ And f ₂ Is Sigmoid function, and the formula is

During the forward propagation of the signal, the output of the hidden layer node is

The output of the node of the output layer is->

/>

And S550, determining error data according to the output power and the preset expected power.

In this scheme, error data may be constructed using the difference between the output power and the preset desired power during the error back propagation.

Specifically, the error data may be E = (E) ₁ ,e ₂ ,L,e _m ) ^T Expressed by an error function of

In the formula e _j Is an error of y _j Is the output power of the neural network, y _j ^* Power is desired for the neural network. And when the error is propagated reversely, calculating the partial derivative of the error vector to the node of the output layer according to a gradient descent method.

S560, if the error data do not meet the preset threshold constraint condition, adjusting parameters in the photovoltaic energy power prediction model until the error data meet the preset threshold constraint condition to obtain an updated photovoltaic energy power prediction model; the parameters comprise connection weights of nodes of a hidden layer and nodes of an output layer in the photovoltaic energy power prediction model and bias parameters of the output layer.

The preset threshold constraint condition may be that the error data is less than or equal to a preset threshold. The size of the preset threshold value can be set according to the photovoltaic energy power prediction requirement.

In this embodiment, when the error data is less than or equal to the preset threshold, the error data meets the preset threshold constraint condition, and at this time, the parameters in the photovoltaic energy power prediction model meet the photovoltaic energy power prediction requirement, so that the parameters do not need to be adjusted. When the error data is larger than the preset threshold, the error data does not meet the constraint condition of the preset threshold, and at the moment, the parameters in the photovoltaic energy power prediction model do not meet the photovoltaic energy power prediction requirement, so that the parameters need to be adjusted.

In particular, the connection weight from the hidden layer node to the output layer node with Sigmoid function is modified to Δ ν _kj ＝αe _j (1-y _j )y _j z _k Correction of the bias parameter of the output layer to DeltaGamma _j ＝αe _j (1-y _j )y _j . Wherein, alpha represents the learning rate, and the value is generally about 1e-3 or less. Too high learning rate may result in non-convergence of model training, and too low learning rate may result in slow model training. β also indicates the learning rate, and may take the same value as α. Too high learning rate may result in non-convergence of model training, and too low learning rate may result in slow model training.

Further, for a single hidden layer structure, the weight of the connection from the input layer node to the hidden layer node is modified to

Correction of bias parameters of a hidden layer

Wherein the determination of the number of hidden layers requires a balance between fitting accuracy and training time. When the number of hidden layers is large, the neural network model has strong generalization capability and high prediction precision. But this consumes more training time. The determination of the number of nodes of the hidden layer also has a great influence on the prediction accuracy. When the number of hidden layer nodes is too small, the training times are required to be increased, and the prediction accuracy is low. When the number of hidden layer nodes is too large, training time is increased, and an overfitting situation is easy to occur at the moment. As the number of hidden layer nodes increases, the actual prediction accuracy generally increases and then decreases.

Furthermore, the cloud computing center tries to reversely propagate the hidden layer node number and the hidden layer number of the neural network by a trial and error method, then uses the test set to check the generalization, and selects a group with the highest generalization. Fig. 6 is a flowchart of a training process of a back propagation neural network according to a third embodiment of the present invention, and as shown in fig. 6, in a model initialization stage, a model randomly sets the number of hidden layers and the number of nodes, connection weights and bias values within a limited numerical range. And estimating the data in the test set by using the trained model at the end stage, checking the sum of the square root and the mean root of the error, recording, and finally selecting one of the multiple groups of models with the smallest sum as a final photovoltaic energy power prediction model.

According to the technical scheme of the embodiment of the invention, the operation historical data of each distributed solar power station is obtained, then each data in the historical data is combined pairwise to obtain the target data, and the historical data and the target data are processed to obtain the characteristics of the training sample. Inputting the characteristics of the training samples into a photovoltaic energy power prediction model to obtain output power, determining error data according to the output power and preset expected power, and if the error data does not meet a preset threshold constraint condition, adjusting parameters in the photovoltaic energy power prediction model until the error data meets the preset threshold constraint condition to obtain an updated photovoltaic energy power prediction model. By executing the technical scheme, the original features can be combined into new features, and then the model is trained, so that the high repetition rate feature training is possible, and meanwhile, the accuracy of photovoltaic energy power prediction can be improved. The multiple edge power prediction platforms realize local prediction of power, on one hand, the trained models only need to be operated, the calculation pressure is low, and on the other hand, the local prediction avoids the situation that the cloud end is attacked by a network, and all photovoltaic power stations cannot predict the power result.

Example four

Fig. 7 is a schematic structural diagram of a photovoltaic energy power prediction apparatus according to a fourth embodiment of the present invention. As shown in fig. 7, the apparatus includes:

the historical data acquisition module 710 is used for acquiring the operating historical data of each distributed solar power station; wherein the historical data consists of power data and meteorological data;

the data combination module 720 is used for combining each data in the historical data in pairs to obtain target data;

a training sample feature obtaining module 730, configured to process the historical data and the target data to obtain training sample features;

and the photovoltaic energy power prediction module 740 is configured to adjust the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model, so as to predict the photovoltaic energy power.

Optionally, the data combining module 720 includes:

the data operation unit is used for performing pairwise operation on each data in the historical data to obtain target data; wherein the operation comprises at least one of addition, subtraction, multiplication, and division.

Optionally, the data operation unit is specifically configured to:

Optionally, the data operation unit is further configured to:

determining the order of each data in the historical data;

Optionally, the training sample feature obtaining module 730 is specifically configured to:

preprocessing the historical data and the target data to obtain the characteristics of the training sample to be processed;

carrying out correlation analysis on the characteristics of the training sample to be processed to obtain a correlation coefficient;

and screening the characteristics of the training sample to be processed according to the correlation coefficient to obtain the characteristics of the training sample.

Optionally, the photovoltaic energy power prediction module 740 is specifically configured to:

inputting the training sample characteristics into a photovoltaic energy power prediction model to obtain output power;

determining error data according to the output power and a preset expected power;

if the error data do not meet the preset threshold constraint condition, adjusting parameters in the photovoltaic energy power prediction model until the error data meet the preset threshold constraint condition to obtain an updated photovoltaic energy power prediction model; the parameters comprise connection weights of nodes of a hidden layer and nodes of an output layer in the photovoltaic energy power prediction model and bias parameters of the output layer.

Optionally, the historical data obtaining module 710 is specifically configured to:

The photovoltaic energy power prediction device provided by the embodiment of the invention can execute the photovoltaic energy power prediction method provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method.

EXAMPLE five

FIG. 8 illustrates a schematic diagram of an electronic device 10 that may be used to implement an embodiment of the invention. Electronic devices are 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 assistants, cellular phones, smart phones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.

As shown in fig. 8, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, and the like, wherein the memory stores a computer program executable by the at least one processor, and the processor 11 can perform various suitable actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from a storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data necessary for the operation of the electronic apparatus 10 may also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to the bus 14.

A number of components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, or the like; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, or the like. Processor 11 performs the various methods and processes described above, such as a photovoltaic energy power prediction method.

In some embodiments, a photovoltaic energy power prediction method may be implemented as a computer program tangibly embodied on a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of a method for photovoltaic energy power prediction as described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform a photovoltaic energy power prediction method by any other suitable means (e.g. by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

A computer program for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be performed. A computer program can execute entirely on a machine, partly on a machine, as a stand-alone software package partly on a machine and partly on a remote machine or entirely on a remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.

The computing system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical host and VPS service are overcome.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present invention may be executed in parallel, sequentially, or in different orders, and are not limited herein as long as the desired result of the technical solution of the present invention can be achieved.

The above-described embodiments should not be construed as limiting the scope of the invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A photovoltaic energy power prediction method is characterized by comprising the following steps:

combining every two data in the historical data to obtain target data;

2. The method of claim 1, wherein combining each data in the historical data two by two to obtain target data comprises:

performing pairwise operation on each data in the historical data to obtain target data; wherein the operation comprises at least one of addition, subtraction, multiplication, and division.

3. The method of claim 2, wherein pairwise computing each data in the historical data to obtain target data comprises:

4. The method of claim 2, wherein pairwise operation is performed on each data in the historical data to obtain target data, further comprising:

determining the order of each data in the historical data;

5. The method of claim 1, wherein processing the historical data and the target data to obtain training sample features comprises:

6. The method of claim 1, wherein adjusting the photovoltaic energy power prediction model according to the training sample characteristics to obtain an updated photovoltaic energy power prediction model for photovoltaic energy power prediction comprises:

7. The method of claim 1, wherein obtaining historical data of operation of each distributed solar power plant comprises:

8. A photovoltaic energy power prediction apparatus, comprising:

9. An electronic device, characterized in that the electronic device comprises:

at least one processor; and

the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform a method of photovoltaic energy power prediction as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon computer instructions for causing a processor to perform, when executed, a method of photovoltaic energy power prediction as claimed in any one of claims 1 to 7.