CN114118596A

CN114118596A - Photovoltaic power generation capacity prediction method and device

Info

Publication number: CN114118596A
Application number: CN202111447540.0A
Authority: CN
Inventors: 万月; 吕贝; 张明杰; 王献文; 孟欣; 梁勇; 易立坤; 王学平; 鲁谟尔; 赵广赫
Original assignee: Huaneng Qingneng Tongyu Electric Power Co ltd; Huaneng Clean Energy Research Institute; Huaneng Longdong Energy Co Ltd
Current assignee: Huaneng Qingneng Tongyu Electric Power Co ltd; Huaneng Clean Energy Research Institute; Huaneng Longdong Energy Co Ltd
Priority date: 2021-11-30
Filing date: 2021-11-30
Publication date: 2022-03-01

Abstract

The application provides a photovoltaic power generation capacity prediction method and device, wherein the method comprises the following steps: the method comprises the steps of obtaining meteorological data and weather types at a plurality of target collection moments, carrying out weather clustering on the meteorological data at the plurality of target collection moments according to the weather types at the plurality of target collection moments to obtain clustered data, and inputting the clustered data into a pre-trained photovoltaic power generation capacity prediction model based on ant colony algorithm optimization to obtain predicted photovoltaic power generation capacity corresponding to the clustered data. According to the photovoltaic power generation prediction method and device, the photovoltaic power generation prediction is carried out on the basis of the clustered data and the ant colony algorithm optimization-based photovoltaic power generation prediction model, the prediction accuracy of the photovoltaic power generation is improved, and the model input is simple.

Description

Photovoltaic power generation capacity prediction method and device

Technical Field

The application relates to the technical field of photovoltaic power generation, in particular to a photovoltaic power generation amount prediction method and device.

Background

The demand of modern society for energy is continuously increasing, and the global energy crisis is increasingly serious. Solar energy plays an increasingly important role and research has shown that the earth instantaneously receives about 1.8 x 1011 mw of energy from solar radiation. The solar photovoltaic power generation system converts solar energy into electric energy, has wide application prospect and is valued by various countries.

The accurate prediction of the photovoltaic power generation amount has important significance for the dispatching management and the stable control of the power system. For the photovoltaic power station, on one hand, price differences in different periods can be fully utilized, energy storage distribution of the system is adjusted, and operation management efficiency of the photovoltaic power station is improved; on the other hand, the photovoltaic power station can be located according to the predicted all-weather performance of the photovoltaic power station.

At present, photovoltaic power generation is generally predicted based on photovoltaic power generation prediction models, and with continuous efforts of scientific researchers, the prediction accuracy of the photovoltaic power generation prediction models is also improved. However, the input of these photovoltaic power generation amount prediction models is generally complicated, and the prediction accuracy of the photovoltaic power generation amount output by the models is to be further improved.

Disclosure of Invention

In view of the above, the present application provides a method and an apparatus for predicting photovoltaic power generation amount, which are used to solve the problem that the input of the existing photovoltaic power generation amount prediction model is complicated, and improve the prediction accuracy of the photovoltaic power generation amount, and the technical scheme is as follows:

a photovoltaic power generation amount prediction method includes:

acquiring meteorological data and weather types at a plurality of target acquisition moments, wherein the meteorological data at one target acquisition moment refers to the meteorological data acquired by a target meteorological factor at the target acquisition moment, and the target meteorological factor is one or more meteorological factors with highest correlation degree with photovoltaic power generation amount in preset meteorological factors;

weather clustering is carried out on the meteorological data at the multiple target acquisition moments according to the weather types at the multiple target acquisition moments to obtain clustered data;

and inputting the clustered data into a pre-trained photovoltaic power generation capacity prediction model optimized based on the ant colony algorithm to obtain predicted photovoltaic power generation capacity corresponding to the clustered data, wherein the photovoltaic power generation capacity prediction model optimized based on the ant colony algorithm is obtained by training with historical clustered data as training samples and historical photovoltaic power generation capacity corresponding to labeled historical clustered data as sample labels.

Optionally, weather clustering is performed on meteorological data at multiple target acquisition moments according to weather types at multiple target acquisition moments to obtain clustered data, including:

determining weather categories respectively corresponding to the weather types at a plurality of target acquisition moments as target weather categories according to preset weather clustering information, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm;

and weather clustering is carried out on the meteorological data at the target acquisition moments according to the target weather category to obtain clustered data under the target weather category.

Optionally, the photovoltaic power generation prediction model optimized based on the ant colony algorithm corresponds to the weather categories in the preset weather clustering information one to one;

inputting the clustered data into a pre-trained photovoltaic power generation capacity prediction model based on ant colony algorithm optimization, wherein the photovoltaic power generation capacity prediction model comprises the following steps:

and respectively inputting the clustered data under the target weather category into the ant colony algorithm optimization-based photovoltaic power generation prediction models corresponding to the target weather category to obtain the predicted photovoltaic power generation corresponding to the clustered data under the target weather category.

Optionally, the method for constructing the photovoltaic power generation capacity prediction model based on ant colony algorithm optimization includes:

acquiring sample data, wherein the sample data comprises historical meteorological data, historical weather types and historical power generation amount data at a plurality of historical acquisition moments;

screening abnormal data in the sample data by adopting a clustering method based on cosine coefficients to obtain the sample data after the abnormal data is screened;

according to the historical weather type in the sample data from which the abnormal data are screened, performing weather clustering on the sample data from which the abnormal data are screened to obtain clustered sample data;

and training to obtain the photovoltaic power generation prediction model based on ant colony algorithm optimization by taking historical meteorological data in the clustered sample data as training input of the BP neural network based on ant colony algorithm optimization and taking historical power generation data in the clustered sample data as training output of the BP neural network based on ant colony algorithm optimization.

Optionally, the method for determining the target weather factor includes:

acquiring meteorological data respectively acquired aiming at each meteorological factor in preset meteorological factors, wherein the meteorological data acquired aiming at the meteorological factor in the preset meteorological factors are meteorological data acquired aiming at the meteorological factors at a plurality of preset acquisition moments;

acquiring historical generated energy data at a plurality of preset acquisition moments;

for each meteorological factor in the preset meteorological factors, calculating the correlation degree of the meteorological data acquired by the meteorological factor and historical power generation amount data at a plurality of preset acquisition moments as the correlation degree corresponding to the meteorological factor to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree of the meteorological factor and the photovoltaic power generation amount;

and sequencing the relevance degrees corresponding to the meteorological factors in the preset meteorological factors respectively so as to determine the target meteorological factor based on the sequencing result.

A photovoltaic power generation amount prediction apparatus comprising: the system comprises an information acquisition module, a meteorological data clustering module and a model application module;

the information acquisition module is used for acquiring meteorological data and weather types at a plurality of target acquisition moments, wherein the meteorological data at one target acquisition moment refers to the meteorological data acquired by a target meteorological factor at the target acquisition moment, and the target meteorological factor is one or more meteorological factors with the highest correlation degree with photovoltaic power generation amount in preset meteorological factors;

the weather data clustering module is used for carrying out weather clustering on the weather data at the plurality of target acquisition moments according to the weather types at the plurality of target acquisition moments to obtain clustered data;

and the model application module is used for inputting the clustered data into a pre-trained ant colony algorithm optimization-based photovoltaic power generation prediction model to obtain predicted photovoltaic power generation corresponding to the clustered data, wherein the ant colony algorithm optimization-based photovoltaic power generation prediction model is obtained by training with historical clustered data as training samples and with historical photovoltaic power generation corresponding to labeled historical clustered data as sample labels.

Optionally, the meteorological data clustering module includes: a target weather category determining submodule and a weather data clustering submodule;

the target weather category determining submodule is used for determining weather categories corresponding to the weather types at a plurality of target acquisition moments respectively according to preset weather clustering information, and the weather categories are used as target weather categories, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm;

and the meteorological data clustering submodule is used for carrying out weather clustering on the meteorological data at the target acquisition moments according to the target weather category to obtain clustered data under the target weather category.

and the model application module is specifically used for respectively inputting the clustered data under the target weather category into the ant colony algorithm optimization-based photovoltaic power generation prediction models corresponding to the target weather category so as to obtain the predicted photovoltaic power generation corresponding to the clustered data under the target weather category.

Optionally, the process of constructing the photovoltaic power generation amount prediction model based on ant colony optimization by the model application module includes: the system comprises a sample data acquisition sub-module, an abnormal data screening sub-module, a sample data clustering sub-module and a model training sub-module;

the sample data acquisition submodule is used for acquiring sample data, wherein the sample data comprises historical meteorological data, historical weather types and historical power generation amount data at a plurality of historical acquisition moments;

the abnormal data screening submodule is used for screening the abnormal data in the sample data by adopting a clustering method based on cosine coefficients to obtain the sample data after the abnormal data is screened;

the sample data clustering submodule is used for carrying out weather clustering on the sample data subjected to abnormal data screening according to the historical weather type in the sample data subjected to abnormal data screening to obtain clustered sample data;

and the model training submodule is used for taking historical meteorological data in the clustered sample data as training input of the BP neural network optimized based on the ant colony algorithm, taking historical power generation data in the clustered sample data as training output of the BP neural network optimized based on the ant colony algorithm, and training to obtain the photovoltaic power generation prediction model optimized based on the ant colony algorithm.

Optionally, the process of determining the target weather factor by the information obtaining module includes: the system comprises a meteorological data acquisition module, a historical power generation data acquisition module, a relevance calculation module and a relevance sequencing module;

the meteorological data acquisition module is used for acquiring meteorological data acquired by aiming at each meteorological factor in preset meteorological factors respectively, wherein the meteorological data acquired by aiming at the meteorological factor in the preset meteorological factors are meteorological data acquired by aiming at the meteorological factors at a plurality of preset acquisition moments;

the historical generating capacity data acquisition module is used for acquiring historical generating capacity data at a plurality of preset acquisition moments;

the correlation degree calculation module is used for calculating the correlation degree of the meteorological data acquired by the meteorological factor and historical power generation amount data at a plurality of preset acquisition moments as the correlation degree corresponding to the meteorological factor so as to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree of the meteorological factor and the photovoltaic power generation amount;

and the association degree sequencing module is used for sequencing the association degrees corresponding to the meteorological factors in the preset meteorological factors respectively so as to determine the target meteorological factors based on the sequencing result.

According to the technical scheme, the photovoltaic power generation prediction method comprises the steps of firstly obtaining meteorological data and weather types at a plurality of target collection moments, then carrying out weather clustering on the meteorological data at the plurality of target collection moments according to the weather types at the plurality of target collection moments to obtain clustered data, and finally inputting the clustered data into a pre-trained photovoltaic power generation prediction model based on ant colony optimization to obtain predicted photovoltaic power corresponding to the clustered data. According to the method and the device, after the influence degree of each meteorological factor on the photovoltaic power generation amount is considered, only meteorological data corresponding to a target meteorological factor with high correlation degree with the photovoltaic power generation amount can be obtained, further, the influence degree of the target meteorological factor on the photovoltaic power generation amount under different weather types is also considered, therefore, the obtained meteorological data can be subjected to weather clustering, the clustered data can be input into a photovoltaic power generation amount prediction model optimized based on an ant colony algorithm to perform photovoltaic power generation amount prediction, on one hand, the photovoltaic power prediction is performed based on the clustered data, the prediction precision of the photovoltaic power generation amount is improved, on the other hand, the photovoltaic power generation amount prediction model optimized based on the ant colony algorithm is adopted, the prediction precision of the photovoltaic power generation amount is further improved, and the model input is simple.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.

Fig. 1 is a schematic flow chart of a photovoltaic power generation amount prediction method according to an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of a hierarchical graph obtained by clustering meteorological samples according to an embodiment of the present disclosure;

fig. 3 is a schematic diagram of a three-layer BP neural network structure provided in an embodiment of the present application;

FIG. 4 is a schematic diagram of anomaly data provided in an embodiment of the present application;

fig. 5 is a scatter diagram of actual solar irradiance values and photovoltaic power generation amounts under 4 weather categories provided by an embodiment of the present application;

fig. 6 is a schematic structural diagram of a photovoltaic power generation amount prediction apparatus according to an embodiment of the present application;

fig. 7 is a block diagram of a hardware structure of a photovoltaic power generation amount prediction apparatus according to an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, 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 application.

The present application provides a method for predicting photovoltaic power generation, which is described in detail below with reference to the following embodiments.

Referring to fig. 1, a schematic flow chart of a photovoltaic power generation amount prediction method provided in an embodiment of the present application is shown, where the photovoltaic power generation amount prediction method may include:

step S101, acquiring meteorological data and weather types at a plurality of target acquisition moments.

The meteorological data at a target acquisition moment refers to meteorological data acquired by a target meteorological factor at the target acquisition moment, and the target meteorological factor is one or more meteorological factors with highest correlation degree with photovoltaic power generation amount in preset meteorological factors.

Optionally, the preset meteorological factors may include, but are not limited to, the following meteorological factors: solar irradiance, clear sky index, maximum air temperature, minimum air temperature, average wind speed, air humidity and air pressure.

It is understood that the meteorological data corresponding to the meteorological factors have an influence on the photovoltaic power generation, for example, for the solar irradiance, the stronger the solar radiation is, i.e., the larger the solar irradiance is, the more the photovoltaic power generation is; for the highest air temperature, the lowest air temperature and the average air temperature, when the air temperature is higher, the open-circuit voltage and the output power of the silicon battery are greatly reduced along with the rise of the air temperature, so that the performance of the silicon battery is reduced, namely the photovoltaic power generation amount is reduced; for the average wind speed, the wind speed can influence the output power by influencing the temperature of the silicon solar cell, and meanwhile, the wind can blow down dust on the surface of the photovoltaic module, so that the transmittance of glass is improved, the output power is improved, and the photovoltaic power generation capacity is increased; in the case of air humidity, an increase in relative humidity means an increase in the water content of the air, which weakens the solar radiation and reduces the output power, i.e. the photovoltaic power generation.

The influence degree of the meteorological factor on the photovoltaic power generation amount can be represented by the correlation degree of the meteorological factor and the photovoltaic power generation amount, wherein the higher the correlation degree of the meteorological factor and the photovoltaic power generation amount is, the more the meteorological data corresponding to the meteorological factor can influence the photovoltaic power generation amount. In this step, one or more weather factors having the highest correlation with the photovoltaic power generation amount among the preset weather factors may be used as the target weather factor. Based on this, this step can acquire meteorological data under a plurality of target collection moments, namely, this step can acquire meteorological data that the target meteorological factor was gathered respectively at every target collection moment. For example, if the target meteorological factors are solar irradiance, average air temperature and air humidity, then taking any target collection time as an example, meteorological data corresponding to solar irradiance, meteorological data corresponding to average air temperature and meteorological data corresponding to air humidity can be collected at the target collection time, and the collected meteorological data is meteorological data at the target collection time in this step.

In an optional embodiment, the step can collect meteorological data under a plurality of target collection moments by the step; preferably, after the meteorological data acquisition platform or the like acquires the meteorological data at a plurality of target acquisition times, the step may directly acquire the meteorological data acquired by the meteorological data acquisition platform, for example, the step may directly acquire the meteorological data acquired by the NASA website at the plurality of target acquisition times.

Furthermore, considering that the weather types represent the time-space distribution of various weather factors in time and space, each weather factor shows different change rules under different weather types, so that the difference of the photovoltaic output power under different weather types is large, namely the influence degree of the target weather factors under different weather types on the photovoltaic power generation amount is different, and the weather types under a plurality of target acquisition moments can be obtained in the step. Optionally, in this embodiment of the application, all weather types may be divided into 13 weather types, such as sunny, cloudy, rainy, snowy, cloudy-cloudy, cloudy-rainy, rainstorm, snow-covered rain, cloudy-rainy, and haze.

Optionally, the step can record weather types at a plurality of target acquisition moments; preferably, the collected weather types can be directly obtained in the step after the weather type collection platform and the like collect the weather data at a plurality of target collection moments.

In this step, in order to make the photovoltaic power generation amount predicted by the present application more accurate, the target collection times may be a plurality of collection times closest to the current time, and of course, the target collection times may be other collection times, which is not limited by the present application.

And S102, carrying out weather clustering on the meteorological data at the multiple target acquisition moments according to the weather types at the multiple target acquisition moments to obtain clustered data.

Optionally, the meteorological data with the same weather type can be classified into one category in this step, so as to obtain clustered data. For example, the target collection times include times t1 to t5, where the weather types at the times t1, t3, and t4 are sunny, and the weather types at the times t2 and t5 are rain, the step may classify the weather data at the times t1, t3, and t4 into one category, and classify the weather data at the times t2 and t5 into one category.

Preferably, considering that the influence degrees of the target meteorological factors on the photovoltaic power generation amount are not much different under some weather types, for example, the influence degrees of the target meteorological factors on the photovoltaic power generation amount are not much different under sunny conditions and cloudy conditions, in order to reduce the number of the weather types and eliminate the influence of different weather types on the prediction accuracy of the model, the weather types can be classified. Based on this, all weather types can be clustered into a plurality of weather categories through a hierarchical clustering algorithm to obtain preset weather clustering information, for example, as shown in table 1 and fig. 2, taking the above 13 weather types as an example, 13 different weather samples are randomly selected and clustered by using the hierarchical clustering algorithm, and in the obtained hierarchical diagram shown in fig. 2, the 13 weather types are clustered into 7 weather categories at most based on the difference of inter-class distances, and when the inter-class distance is 3, the 13 weather types are clustered into 4 weather categories, that is, the weather clustering information shown in table 1 below.

TABLE 1 weather clustering information

It should be noted that, in the present application, a hierarchical clustering method is adopted to obtain a correlation coefficient r between the clear sky index and the solar radiation amount under each meteorological sample, and a linear slope k value obtained by a fitting method, and the linear slope k value is used as a classification variable of the hierarchical clustering, and the euclidean distance of each meteorological sample is used as an evaluation standard to perform cluster analysis on the weather types.

Here, the expression of euclidean distance is:

where a, b represent the selected classification variables and n represents the classification dimension of the variables.

The hierarchical clustering method mainly comprises the following steps: transforming the sample group; converting the processed object into n classes, each class containing a sample; calculating the distance between each two levels according to the measured values; merging the two most recent classes into a new class; and calculating the distance between the new class and the current class, combining the two classes with the closer distance into one class, and continuously operating the steps.

Based on the above description, the specific implementation process of this step may include the following steps S1021 to S1022:

and S1021, determining weather categories corresponding to the weather types at a plurality of target acquisition moments respectively according to preset weather clustering information, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm.

In this step, after the weather types at the multiple target collection times are obtained, the obtained weather types may be clustered according to the preset weather clustering information.

For example, the multiple target collection times include times t1 to t5, where the weather type at time t1 is clear-to-cloudy, the weather type at time t2 is gusty rain, the weather types at times t3 and t4 are clear, and the weather type at time t5 is rain, in this step, the weather types at times t1, t3, and t4 are all a, and the weather types at times t2 and t5 are all D.

Step S1022, weather clustering is carried out on the meteorological data at the target collection moments according to the target weather category, and clustered data under the target weather category are obtained.

For example, based on the example shown in the previous step, the present step can classify the meteorological data at times t1, t3, and t4 into one category, and classify the meteorological data at times t2 and t5 into one category.

And S103, inputting the clustered data into a pre-trained photovoltaic power generation amount prediction model based on ant colony optimization so as to obtain the predicted photovoltaic power generation amount corresponding to the clustered data.

The photovoltaic power generation prediction model based on ant colony algorithm optimization is obtained by training by taking historical clustered data as a training sample and taking historical photovoltaic power generation corresponding to labeled historical clustered data as a sample label.

It should be noted that the present step is implemented in various ways, and the present application provides, but is not limited to, the following two ways.

The first implementation mode comprises the following steps: optionally, in this step, an ant colony algorithm optimization-based photovoltaic power generation amount prediction model may be trained for the clustered data, in the ant colony algorithm optimization-based photovoltaic power generation amount prediction model, photovoltaic power generation amount prediction may be performed based on various data in the clustered data, and then, predicted photovoltaic power generation amounts corresponding to the various data are superimposed to obtain a total predicted photovoltaic power generation amount. For example, if the meteorological data at the times t1, t3 and t4 in the clustered data are data of one type, and the meteorological data at the times t2 and t5 are data of one type, the clustered data are input into a photovoltaic power generation amount prediction model optimized based on the ant colony algorithm in the step, the model can predict a photovoltaic power generation amount based on the meteorological data at the times t1, t3 and t4, predict a photovoltaic power generation amount based on the meteorological data at the times t2 and t5, and the predicted total photovoltaic power generation amount obtained by superposing the two photovoltaic power generation amounts is the predicted photovoltaic power generation amount corresponding to the clustered data output by the model.

The second implementation mode comprises the following steps: optionally, it can be understood that, when photovoltaic power generation amount prediction is performed, a weather category which has a large influence on photovoltaic power generation amount and a target meteorological factor under the weather category are mainly considered. Based on this, in this step, an ant colony algorithm optimization-based photovoltaic power generation amount prediction model may be trained for each type of data in the clustered data, that is, an ant colony algorithm optimization-based photovoltaic power generation amount prediction model may be trained for each weather category (e.g., A, B, C and D in step S102), that is, the ant colony algorithm optimization-based photovoltaic power generation amount prediction model corresponds to the weather category in the preset weather clustering information. At this time, the step of inputting the clustered data into the pre-trained ant colony algorithm optimization-based photovoltaic power generation prediction model specifically includes: and respectively inputting the clustered data under the target weather category into the ant colony algorithm optimization-based photovoltaic power generation prediction models corresponding to the target weather category to obtain the predicted photovoltaic power generation corresponding to the clustered data under the target weather category.

For example, if the meteorological data at the time t1, t3 and t4 in the clustered data are of a type (the weather category corresponding to the data is a), and the meteorological data at the time t2 and t5 are of a type (the weather category corresponding to the data is D), the target weather category is a and D, in this step, the meteorological data at the time t1, t3 and t4 can be input to the photovoltaic power generation amount prediction model corresponding to the weather category a and optimized based on the ant colony optimization, so as to obtain the predicted photovoltaic power generation amount corresponding to the meteorological data at the time t1, t3 and t4 output by the model; in addition, in this step, the meteorological data at the time t2 and the time t5 may be input to the ant colony optimization-based photovoltaic power generation amount prediction model corresponding to the weather category D, so as to obtain the predicted photovoltaic power generation amount corresponding to the meteorological data at the time t2 and the time t5, which are output by the model.

It should be further noted that the photovoltaic power generation amount prediction model optimized based on the ant colony algorithm in this step may be a BP neural network model optimized based on the ant colony algorithm. In order to make the person skilled in the art understand the ant colony algorithm optimization-based BP neural network model, the following detailed description is provided.

Neural networks or artificial neural networks refer to nonlinear systems, i.e., neurons, composed of a large number of simple computational elements. It imitates the information processing, storing and searching function of human brain nervous system to a certain extent, and is a kind of simplification, abstraction and simulation for human brain. Neural networks are composed of a large number of interconnected neurons. The BP neural network is a feedforward network, and is the most common neural network algorithm. It uses the output error to estimate the error of the output layer and then uses this error to estimate the error of the previous layer. Thus, error estimates for all other layers can be obtained by layer-by-layer retransmission. The main idea of the BP neural network algorithm is as follows: learning samples' x for N inputs₁，x₂，...x_n'corresponding M output samples are' t₁，t₂，...，t_m'use of the error between the actual outputs of the network' z₁，z₂，...，z_m' and the target vector't 1, t 2.., tm ' to modify the weights so that the output z of the network_l(l 1, 2.. m) is close to the desired value t_l. A typical layer 3 BP neuron network is shown in fig. 3. In fig. 3, the Input layer of the network is x ═ x (x)₁，x₂，...，x_n)^TA hidden layerThe Hidden layer has h element, and the output of the Hidden layer is y ═ y₁，y₂，...，y_h)^TThe Output layer (i.e. Output layer) has m elements, and the Output is z ═ z₁，z₂，...，z_m)^TThe target output is t ═ t (t)₁，t₂，...，t_m)^T. Assuming that the transfer function from the hidden layer to the output layer is f and the transfer function of the output layer is g, it can be obtained

In the formula, y_jIs the output of the jth nervous system in the hidden layer, and w_0j＝θ，x₀Is-1, and can be obtained

In the formula, z_kThe output of the k-th neuron of the output layer is based on the error between the network output and the target output:

all that needs to be done is to adjust the weight values such that the error epsilon is minimized. The process of network training is to find appropriate weights and biases to minimize the cost function. In the error back propagation process, the main work is to calculate the gradient of the parameter according to a cost function formed by the difference between the prediction result and the actual result, and update the weight and the deviation by using a gradient descent method so as to reduce the cost function as soon as possible, thereby completing the network training. Because the initial weight and the deviation of a single BP neural network are completely random, although the weight and the deviation can be corrected by the BP network in the training process, the initial weight and the deviation have obvious influence on the result, if the initial value is not properly selected, the convergence speed of the network can be slowed down or slowed down, or a local minimum solution rather than a global optimal solution can be obtained. Intelligent algorithms can be used to optimize the training of the BP neural network to solve this problem.

In the application, after various intelligent algorithms are comprehensively considered, the ant colony algorithm is determined to be adopted to optimize the initial weight of the BP neural network. The ant colony Algorithm (ACO) is a bionic algorithm based on simulating ant path finding in nature. Ants can leave pheromones on the passing path during movement to transmit information, and can sense the substances and guide the movement direction of the substances during movement. Therefore, the collective behavior of a colony consisting of a large number of ants exhibits a positive feedback phenomenon: the more ants walking on a certain path, the greater the probability of selecting a path later. The ant colony algorithm has intelligent global optimization capability and good robustness, and the weight of the BP neural network is trained by the ant colony algorithm, so that a better initial weight can be obtained.

On the basis of the above description, the method for constructing the photovoltaic power generation amount prediction model based on ant colony optimization according to the present application may include the following steps S1031 to S1034:

and step S1031, obtaining sample data, wherein the sample data comprises historical meteorological data, historical weather types and historical power generation amount data at a plurality of historical acquisition moments, and the historical meteorological data at one historical acquisition moment refers to the historical meteorological data acquired by aiming at the target meteorological factor at the historical acquisition moment.

For example, the target weather factor is solar irradiance, and the solar irradiance values collected for the solar irradiance on the first day of each month in 2020 are h₁～h₁₂The weather type of the first day of each month in 2020 is s₁～s₁₂Historical power generation data is p₁～p₁₂If the historical meteorological data at the multiple historical acquisition moments obtained in this step is h₁～h₁₂Historical weather type is s₁～s₁₂Historical power generation data is p₁～p₁₂。

And S1032, screening abnormal data in the sample data by adopting a clustering method based on the cosine coefficient to obtain the sample data with the abnormal data screened.

It is understood that there may be some cases where abnormal data is collected at some historical collection time, for example, due to temporary shutdown, equipment failure, or snow accumulation, etc., which may result in abnormal data being collected at individual historical collection time, i.e., the data characteristics collected at individual historical collection time are different from the main sample in the sample data. In order to optimize the dispersibility of sample data and improve the prediction accuracy, a clustering method using a cosine coefficient as an evaluation standard can be adopted to screen abnormal data in the sample data so as to screen the abnormal data acquired at individual historical acquisition moments. For example, if the target meteorological factor includes solar irradiance, the step may use the photovoltaic power generation amount and the solar irradiance as a standard, and use a clustering method using a cosine coefficient as an evaluation standard to screen the abnormal data. Specifically, for each historical acquisition time corresponding to the sample data, the cosine distance between the solar irradiance value and the historical power generation amount data in the historical meteorological data at the historical acquisition time is calculated by adopting the following cosine distance calculation formula, if the calculated cosine distance is within the preset distance threshold range, the historical meteorological data, the historical weather type and the historical power generation amount data at the historical acquisition time are valid data, and if not, the historical meteorological data, the historical weather type and the historical power generation amount data at the historical acquisition time are abnormal data.

The formula for calculating the cosine distance is as follows:

in the formula u_iAnd v_iThe historical all-day solar irradiance value and the historical all-day photovoltaic power generation amount at the historical collection time i are respectively.

For example, the sample data obtained in step S1031 includes a history acquisition time t'₁～t'₁₂The abnormal data screened out in the step are historical meteorological data, historical weather types and historical power generation amount data, and the abnormal data screened out in the step are historical acquisition time t'₁And t'₁₂And if the historical meteorological data, the historical weather type and the historical power generation data are obtained, the sample data obtained in the step and subjected to abnormal data screening comprises historical acquisition time t'₂～t'₁₁Historical meteorological data, historical weather types and historical power generation data.

And S1033, performing weather clustering on the sample data with the abnormal data removed according to the historical weather type in the sample data with the abnormal data removed to obtain clustered sample data.

The specific implementation process of this step corresponds to the specific implementation process of step S102, and reference may be made to the description in the foregoing step for details, which are not repeated herein.

And S1034, taking historical meteorological data in the clustered sample data as training input of a BP neural network optimized based on the ant colony algorithm, taking historical power generation data in the clustered sample data as training output of the BP neural network optimized based on the ant colony algorithm, and training to obtain a photovoltaic power generation prediction model optimized based on the ant colony algorithm.

Specifically, the method comprises the following implementation steps:

and Step1, acquiring historical meteorological data in the clustered sample data as training input of the BP neural network optimized based on the ant colony algorithm, and acquiring historical power generation data in the clustered sample data as training output of the BP neural network optimized based on the ant colony algorithm.

Step 2: and initializing the weight value and the threshold length of the BP neural network, and determining the initial parameters of the ant colony algorithm. The number of ants M, the maximum iteration number T, the concentration of pheromone, the evaporation coefficient of pheromone and the like.

Step 3: and (5) placing the M ants at an initial position, and obtaining and recording the optimal path by using an ant colony algorithm. And updating the search taboo table and the pheromone table.

Step 4: and before the final condition is met, assigning the optimal weight obtained after optimization to the BP neural network, and establishing a BP neural network model. And training the BP neural network model by using the data obtained at Step1 until the error meets the requirement.

According to the photovoltaic power generation capacity prediction method, firstly, meteorological data and weather types at a plurality of target collection moments are obtained, then, weather clustering is carried out on the meteorological data at the plurality of target collection moments according to the weather types at the plurality of target collection moments to obtain clustered data, and finally, the clustered data are input into a pre-trained photovoltaic power generation capacity prediction model optimized based on an ant colony algorithm to obtain predicted photovoltaic power corresponding to the clustered data. According to the method and the device, after the influence degree of each meteorological factor on the photovoltaic power generation amount is considered, only meteorological data corresponding to a target meteorological factor with high correlation degree with the photovoltaic power generation amount can be obtained, further, the influence degree of the target meteorological factor on the photovoltaic power generation amount under different weather types is also considered, therefore, the obtained meteorological data can be subjected to weather clustering, the clustered data can be input into a photovoltaic power generation amount prediction model optimized based on an ant colony algorithm to perform photovoltaic power generation amount prediction, on one hand, the photovoltaic power prediction is performed based on the clustered data, the prediction precision of the photovoltaic power generation amount is improved, on the other hand, the photovoltaic power generation amount prediction model optimized based on the ant colony algorithm is adopted, the prediction precision of the photovoltaic power generation amount is further improved, and the model input is simple.

In one embodiment of the present application, a determination process of a target weather factor is described.

Optionally, the process may include the following steps S1 to S4:

step S1, acquiring meteorological data respectively acquired for each meteorological factor in the preset meteorological factors, where the meteorological data acquired for an meteorological factor in the preset meteorological factors are meteorological data acquired for the meteorological factor at a plurality of preset acquisition times.

Optionally, the preset collection time may be a historical collection time. Of course, the preset collection time may be other collection times, which is not limited in the present application.

And step S2, acquiring historical power generation amount data at a plurality of preset acquisition moments.

Step S3, for each meteorological factor in the preset meteorological factors, calculating the correlation degree between the meteorological data acquired by the meteorological factor and the historical power generation amount data at a plurality of preset acquisition moments as the correlation degree corresponding to the meteorological factor so as to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree between the meteorological factor and the photovoltaic power generation amount.

Considering that the number of model inputs and the coupling relation between input variables have important influence on the prediction precision of the model when the neural network model is used for predicting the photovoltaic power generation amount, in order to screen suitable model input variables, the correlation between the photovoltaic power generation amount influence factors is analyzed by adopting a gray correlation analysis (GRA) method, so that the meteorological factor (namely the target meteorological factor) with strong correlation is selected from the preset meteorological factors.

Note that, the calculation formula of the gray correlation coefficient used in this step is as follows.

In the formula, q₀Is a reference sequence (i.e. a sequence consisting of historical generated energy data at a plurality of preset acquisition moments), q_iFor comparison of the sequence (i.e. the sequence of meteorological data collected for meteorological factor i in the preset meteorological factors), ρ is a resolution coefficient, which may be optionally 0.5 in the calculation. Xi_iThe correlation coefficient of the reference sequence and the ith comparison sequence at K is shown, and the similarity of different comparison sequences and the reference sequence at the same point is shown.

The grey correlation calculation formula adopted in the step is as follows.

Wherein n represents the number of meteorological factors contained in the preset meteorological factor, gamma_0iAnd representing the relevance corresponding to the meteorological factor i.

And S4, sequencing the relevance degrees corresponding to the meteorological factors in the preset meteorological factors respectively so as to determine the target meteorological factor based on the sequencing result.

In this step, one or more weather factors with a higher degree of association may be selected from the sorting result as the target weather factor according to the actual situation.

In order that those skilled in the art will better understand the present application, the following description refers to a specific embodiment.

The model training process comprises the following steps 1-5:

step1: and acquiring historical meteorological data acquired by the meteorological factors at a plurality of historical acquisition moments, and acquiring historical power generation data and historical weather types at a plurality of historical acquisition moments.

For example, the preset meteorological factors include the following 8 meteorological factors: solar irradiance, clear sky index, maximum air temperature, minimum air temperature, average wind speed, air humidity and air pressure. Assuming that the plurality of historical acquisition moments includes 693 historical acquisition moments, 693 × 8 historical meteorological data can be obtained in total in this step, and 100 historical power generation data can be obtained in total in this step.

Step 2: and calculating the correlation degree corresponding to each meteorological factor contained in the preset meteorological factor, and screening the meteorological factor with higher correlation degree as the target meteorological factor according to the calculated correlation degree.

The process of this step can refer to the above steps S1 to S4, which are not described herein again.

For example, the target meteorological factors obtained in this step include the following 7 meteorological factors: solar irradiance, clear sky index, maximum air temperature, minimum air temperature, average wind speed and air humidity. Then, in this step, 93 × 7 pieces of historical meteorological data, 693 pieces of historical power generation data at the historical acquisition time, and the historical weather type may be used as sample data.

And step 3: and (3) screening abnormal data in the data obtained in the step (1) by adopting a cosine coefficient-based clustering method.

The process of this step may refer to step S1032, which is not described herein again.

For example, referring to fig. 4, a schematic diagram of the step of screening 78 sets of Abnormal data from 693 sets of Sample data (data acquired at each historical acquisition time is taken as a set) acquired at 693 historical acquisition times is shown, where the Sample data in the shape of a "star" is Abnormal data (Abnormal Sample), and the Sample data in the shape of a "square" is Valid data (Valid Sample), and it is seen that Abnormal data can be well screened in the step.

And 4, step 4: and according to the historical weather type in the sample data from which the abnormal data is removed, performing weather clustering on the sample data from which the abnormal data is removed to obtain the clustered sample data.

The process of this step may refer to step S1033 or step S102, which is not described herein in detail.

For example, when cluster screening is performed on 615 sets of valid sample data from which abnormal data has been screened based on the 4 weather categories provided in S102, as shown in fig. 5, a scatter diagram of the actual Solar irradiance value (Solar Radiation) and the photovoltaic Power Generation amount (Power Generation) shows that the Solar irradiance has different degrees of influence on the photovoltaic Power Generation amount in different weather categories.

And 5: and training to obtain the photovoltaic power generation prediction model based on ant colony algorithm optimization by taking historical meteorological data in the clustered sample data as training input of the BP neural network based on ant colony algorithm optimization and taking historical power generation data in the clustered sample data as training output of the BP neural network based on ant colony algorithm optimization.

The process of this step may refer to step S1034, which is not described herein again.

The model application process comprises the following steps 6-8:

step 6: and acquiring meteorological data and weather types at a plurality of target acquisition moments.

And 7: and according to the weather types at the multiple target acquisition moments, performing weather clustering on the weather data at the multiple target acquisition moments to obtain clustered data.

And 8: and inputting the clustered data into a pre-trained photovoltaic power generation prediction model based on ant colony algorithm optimization to obtain the predicted photovoltaic power generation corresponding to the clustered data.

The processes of step 6 to step 8 correspond to the processes of step S101 to step S103, and reference may be made to the foregoing description for details, which are not repeated herein.

It should be noted that the present embodiment is only an example, and is not a limitation to the present application.

It should be further noted that the predicted photovoltaic power generation amount obtained by the method can be used for operation and maintenance of the photovoltaic power station and can be used as one of judgment references for judging whether the power station operation state is normal or not. In addition, the photovoltaic power generation prediction method and the photovoltaic power generation prediction device can provide basis for site selection of the power station.

The embodiment of the application also provides a photovoltaic power generation amount prediction device, which is described below, and the photovoltaic power generation amount prediction device described below and the photovoltaic power generation amount prediction method described above can be referred to correspondingly.

Referring to fig. 6, a schematic structural diagram of a photovoltaic power generation amount prediction apparatus provided in an embodiment of the present application is shown, and as shown in fig. 6, the photovoltaic power generation amount prediction apparatus may include: the system comprises an information acquisition module 601, a meteorological data clustering module 602 and a model application module 603.

The information acquisition module 601 is configured to acquire meteorological data and weather types at multiple target acquisition moments, where the meteorological data at a target acquisition moment refers to meteorological data acquired by a target meteorological factor at the target acquisition moment, and the target meteorological factor is one or more meteorological factors with the highest correlation degree with photovoltaic power generation amount in preset meteorological factors.

And the meteorological data clustering module 602 is configured to perform weather clustering on meteorological data at multiple target acquisition moments according to weather types at multiple target acquisition moments to obtain clustered data.

The model application module 603 is configured to input the clustered data into a pre-trained ant colony algorithm-optimized photovoltaic power generation prediction model to obtain predicted photovoltaic power generation corresponding to the clustered data, where the ant colony algorithm-optimized photovoltaic power generation prediction model is obtained by training with historical clustered data as a training sample and with a historical photovoltaic power generation corresponding to labeled historical clustered data as a sample label.

The application provides a photovoltaic power generation capacity prediction device, at first acquire meteorological data and weather type under a plurality of target acquisition moments, then according to the weather type under a plurality of target acquisition moments, carry out weather clustering to meteorological data under a plurality of target acquisition moments, obtain the data after clustering, at last with the data after clustering input to the photovoltaic power generation capacity prediction model based on ant colony algorithm optimization trained in advance to obtain the prediction photovoltaic power that the data after clustering corresponds. According to the method and the device, after the influence degree of each meteorological factor on the photovoltaic power generation amount is considered, only meteorological data corresponding to a target meteorological factor with high correlation degree with the photovoltaic power generation amount can be obtained, further, the influence degree of the target meteorological factor on the photovoltaic power generation amount under different weather types is also considered, therefore, the obtained meteorological data can be subjected to weather clustering, the clustered data can be input into a photovoltaic power generation amount prediction model optimized based on an ant colony algorithm to perform photovoltaic power generation amount prediction, on one hand, the photovoltaic power prediction is performed based on the clustered data, the prediction precision of the photovoltaic power generation amount is improved, on the other hand, the photovoltaic power generation amount prediction model optimized based on the ant colony algorithm is adopted, the prediction precision of the photovoltaic power generation amount is further improved, and the model input is simple.

In a possible implementation manner, the meteorological data clustering module 602 may include:

and the target weather category determining submodule is used for determining weather categories corresponding to the weather types at a plurality of target acquisition moments respectively according to preset weather clustering information, and the weather categories are used as target weather categories, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm.

In a possible implementation manner, the photovoltaic power generation amount prediction model optimized based on the ant colony algorithm corresponds to weather categories in preset weather clustering information one to one; based on this, the model application module 603 may be specifically configured to input the clustered data in the target weather category to the ant colony optimization-based photovoltaic power generation amount prediction model corresponding to the target weather category, so as to obtain predicted photovoltaic power generation amounts corresponding to the clustered data in the target weather category.

In a possible implementation manner, the process of constructing the photovoltaic power generation amount prediction model based on ant colony optimization by the model application module 603 may include:

the sample data acquisition submodule is used for acquiring sample data, wherein the sample data comprises historical meteorological data, historical weather types and historical power generation amount data at a plurality of historical acquisition moments, and the historical meteorological data at one historical acquisition moment refers to the historical meteorological data acquired by aiming at the target meteorological factor at the historical acquisition moment.

And the abnormal data screening submodule is used for screening the abnormal data in the sample data by adopting a clustering method based on cosine coefficients to obtain the sample data after the abnormal data is screened.

And the sample data clustering submodule is used for carrying out weather clustering on the sample data after the abnormal data is screened out according to the historical weather type in the sample data after the abnormal data is screened out, so as to obtain the clustered sample data.

In a possible implementation manner, the process of determining the target weather factor by the information obtaining module may include:

and the meteorological data acquisition module is used for acquiring meteorological data acquired aiming at each meteorological factor in the preset meteorological factors respectively, wherein the meteorological data acquired aiming at the meteorological factor in the preset meteorological factors are meteorological data acquired aiming at a plurality of preset acquisition moments acquired by the meteorological factors.

And the historical generating capacity data acquisition module is used for acquiring the historical generating capacity data at the preset acquisition moments.

And the correlation degree calculation module is used for calculating the correlation degree of the meteorological data acquired by the meteorological factor and the historical power generation amount data at the multiple preset acquisition moments as the correlation degree corresponding to the meteorological factor so as to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree of the meteorological factor and the photovoltaic power generation amount.

And the association degree sequencing module is used for sequencing the association degrees respectively corresponding to the meteorological factors in the preset meteorological factors so as to determine the target meteorological factor based on the sequencing result.

The embodiment of the application also provides photovoltaic power generation capacity prediction equipment. Alternatively, fig. 7 shows a block diagram of a hardware structure of the photovoltaic power generation amount prediction apparatus, and referring to fig. 7, the hardware structure of the photovoltaic power generation amount prediction apparatus may include: at least one processor 701, at least one communication interface 702, at least one memory 703 and at least one communication bus 704;

in the embodiment of the present application, the number of the processor 701, the communication interface 702, the memory 703 and the communication bus 704 is at least one, and the processor 701, the communication interface 702 and the memory 703 complete mutual communication through the communication bus 704;

the processor 701 may be a central processing unit CPU, or an application Specific Integrated circuit (asic), or one or more Integrated circuits configured to implement embodiments of the present invention, or the like;

the memory 703 may include a high-speed RAM memory, and may further include a non-volatile memory (non-volatile memory) or the like, such as at least one disk memory;

wherein the memory 703 stores a program, and the processor 701 may call the program stored in the memory 703, the program being configured to:

Alternatively, the detailed function and the extended function of the program may be as described above.

The embodiment of the application also provides a readable storage medium, wherein a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the photovoltaic power generation amount prediction method is realized.

Finally, it is further noted that, herein, relational terms such as, for example, and the like may be 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 phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

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

weather clustering is carried out on the meteorological data at the target acquisition moments according to the weather types at the target acquisition moments to obtain clustered data;

and inputting the clustered data into a pre-trained photovoltaic power generation capacity prediction model based on ant colony algorithm optimization to obtain the predicted photovoltaic power generation capacity corresponding to the clustered data, wherein the photovoltaic power generation capacity prediction model based on ant colony algorithm optimization is obtained by taking historical clustered data as training samples and taking the marked historical photovoltaic power generation capacity corresponding to the historical clustered data as sample labels for training.

2. The photovoltaic power generation capacity prediction method according to claim 1, wherein the weather clustering is performed on the meteorological data at the target collection times according to the weather types at the target collection times to obtain clustered data, and the method comprises:

determining weather categories respectively corresponding to the weather types at the target acquisition moments as target weather categories according to preset weather clustering information, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm;

3. The photovoltaic power generation amount prediction method according to claim 2, wherein the ant colony algorithm-based optimized photovoltaic power generation amount prediction model corresponds to weather categories in the preset weather cluster information one to one;

the step of inputting the clustered data into a pre-trained photovoltaic power generation capacity prediction model based on ant colony optimization comprises the following steps:

and respectively inputting the clustered data under the target weather category to a photovoltaic power generation prediction model which corresponds to the target weather category and is optimized based on an ant colony algorithm, so as to obtain predicted photovoltaic power generation which corresponds to the clustered data under the target weather category respectively.

4. The photovoltaic power generation amount prediction method according to any one of claims 1 to 3, wherein the method for constructing the ant colony algorithm optimization-based photovoltaic power generation amount prediction model comprises the following steps:

according to the historical weather type in the sample data after the abnormal data is screened out, performing weather clustering on the sample data after the abnormal data is screened out to obtain clustered sample data;

and training to obtain the photovoltaic power generation prediction model based on ant colony algorithm optimization by taking the historical meteorological data in the clustered sample data as training input of a BP neural network based on ant colony algorithm optimization and taking the historical power generation data in the clustered sample data as training output of the BP neural network based on ant colony algorithm optimization.

5. The photovoltaic power generation amount prediction method according to claim 4, characterized in that the determination method of the target meteorological factor includes:

acquiring meteorological data respectively acquired aiming at each meteorological factor in the preset meteorological factors, wherein the meteorological data acquired aiming at the meteorological factor in the preset meteorological factors are meteorological data acquired aiming at the meteorological factors at a plurality of preset acquisition moments;

for each meteorological factor in the preset meteorological factors, calculating the correlation degree between the meteorological data acquired by the meteorological factor and the historical power generation amount data at the multiple preset acquisition moments as the correlation degree corresponding to the meteorological factor so as to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree between the meteorological factor and the photovoltaic power generation amount;

and sequencing the relevance degrees respectively corresponding to the meteorological factors in the preset meteorological factors so as to determine the target meteorological factor based on a sequencing result.

6. A photovoltaic power generation amount prediction apparatus characterized by comprising: the system comprises an information acquisition module, a meteorological data clustering module and a model application module;

the meteorological data clustering module is used for carrying out weather clustering on the meteorological data at the target acquisition moments according to the weather types at the target acquisition moments to obtain clustered data;

the model application module is used for inputting the clustered data into a pre-trained ant colony algorithm optimization-based photovoltaic power generation prediction model to obtain predicted photovoltaic power generation corresponding to the clustered data, wherein the ant colony algorithm optimization-based photovoltaic power generation prediction model is obtained by training with historical clustered data as training samples and with marked historical photovoltaic power generation corresponding to the historical clustered data as sample labels.

7. The photovoltaic power generation amount prediction apparatus according to claim 6, wherein the meteorological data clustering module includes: a target weather category determining submodule and a weather data clustering submodule;

the target weather category determining submodule is used for determining weather categories corresponding to the weather types at the target acquisition moments respectively according to preset weather clustering information, and the weather categories are used as target weather categories, wherein the preset weather clustering information is obtained by clustering all the weather types into a plurality of weather categories through a hierarchical clustering algorithm;

8. The photovoltaic power generation amount prediction device according to claim 7, wherein the ant colony algorithm-based optimized photovoltaic power generation amount prediction model corresponds to weather categories in the preset weather cluster information one to one;

the model application module is specifically configured to input the clustered data in the target weather category to the ant colony optimization-based photovoltaic power generation prediction models corresponding to the target weather category, so as to obtain predicted photovoltaic power generation corresponding to the clustered data in the target weather category.

9. The photovoltaic power generation amount prediction device according to any one of claims 6 to 8, wherein the process of constructing the ant colony algorithm-based optimized photovoltaic power generation amount prediction model by the model application module includes: the system comprises a sample data acquisition sub-module, an abnormal data screening sub-module, a sample data clustering sub-module and a model training sub-module;

and the model training submodule is used for training to obtain the photovoltaic power generation prediction model based on ant colony algorithm optimization by taking the historical meteorological data in the clustered sample data as training input of a BP neural network based on ant colony algorithm optimization and taking the historical power generation data in the clustered sample data as training output of the BP neural network based on ant colony algorithm optimization.

10. The photovoltaic power generation amount prediction apparatus according to claim 9, wherein the process of the information acquisition module determining the target meteorological factor includes: the system comprises a meteorological data acquisition module, a historical power generation data acquisition module, a relevance calculation module and a relevance sequencing module;

the meteorological data acquisition module is used for acquiring meteorological data acquired by aiming at each meteorological factor in the preset meteorological factors respectively, wherein the meteorological data acquired by aiming at the meteorological factor in the preset meteorological factors are meteorological data acquired by aiming at the meteorological factor at a plurality of preset acquisition moments;

the historical generating capacity data acquisition module is used for acquiring historical generating capacity data at the preset acquisition moments;

the correlation degree calculating module is used for calculating the correlation degree between the meteorological data acquired by the meteorological factor and the historical power generation amount data at the multiple preset acquisition moments as the correlation degree corresponding to the meteorological factor so as to obtain the correlation degree corresponding to each meteorological factor in the preset meteorological factors, wherein the correlation degree corresponding to the meteorological factor represents the correlation degree between the meteorological factor and the photovoltaic power generation amount;

the relevancy sorting module is configured to sort the relevancy corresponding to each weather factor in the preset weather factors, so as to determine the target weather factor based on a sorting result.