Distributed photovoltaic power station generation power prediction method and device

Abstract

The invention provides a distributed photovoltaic power station generation power prediction method and device, and the method comprises the steps: clustering photovoltaic power stations in a distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of a to-be-predicted photovoltaic power station from the cluster where the to-be-predicted photovoltaic power station is located in the photovoltaic power stations; according to the value of the key meteorological characteristic of the reference photovoltaic power station in the preset time period, obtaining the first predicted power generation power of the reference photovoltaic power station in the preset time period; and according to the value of the key meteorological characteristic of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation power of the reference photovoltaic power station in the preset time period, obtaining the second predicted power generation power of the photovoltaic power station to be predicted in the preset time period. The prediction precision of the generated power of the distributed photovoltaic power station is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method and device for predicting the power generation of a distributed photovoltaic power station. Background Art

[0002] Photovoltaic power generation is widely used as a clean renewable energy source. Photovoltaic power generation has a diurnal periodicity and relies on many meteorological variables, such as solar irradiance, air temperature, cloud changes, wind speed, relative humidity, etc. These meteorological variables are highly uncertain in nature, making photovoltaic power generation intermittent, volatile and random. Therefore, after large-scale grid connection of photovoltaic power stations, the power grid is unstable. It is necessary to predict the power generation of photovoltaic power stations in order to control the power generation of photovoltaic power stations and improve the stability of the power grid.

[0003] Traditional centralized photovoltaic power stations are usually built in specific geographical locations. The power forecast of photovoltaic power stations is usually modeled based on time series, and the data used is in a single form. The accuracy of using this method to predict the power generation of distributed photovoltaic power stations is low. Summary of the invention

[0004] The present invention provides a method and device for predicting the power generation of a distributed photovoltaic power station, which are used to solve the defect of low accuracy in the prediction of the power generation of a distributed photovoltaic power station in the prior art, and to improve the prediction accuracy of the power generation of a distributed photovoltaic power station.

[0005] The present invention provides a method for predicting power generation of a distributed photovoltaic power station, comprising:

[0006] Clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station;

[0007] Obtaining a first predicted power generation of the reference photovoltaic power station within the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station within the preset time period;

[0008] According to the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation of the reference photovoltaic power station in the preset time period, the second predicted power generation of the photovoltaic power station to be predicted in the preset time period is obtained.

[0009] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the photovoltaic power stations in the distributed photovoltaic power station cluster are clustered to obtain multiple clusters, including:

[0010] Selecting the historical power of the first preset date of each season from the annual historical power of the photovoltaic power station;

[0011] Determine one or more of the mean, standard deviation, coefficient of variation, kurtosis and skewness of the daily historical power in the historical power of each first preset date;

[0012] The photovoltaic power station is clustered to obtain a plurality of clusters according to one or more of the mean value, standard deviation, coefficient of variation, kurtosis and skewness.

[0013] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the photovoltaic power stations in the distributed photovoltaic power station cluster are clustered to obtain multiple clusters, and a reference photovoltaic power station of the photovoltaic power station to be predicted is selected from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station, including:

[0014] Clustering the photovoltaic power stations in the distributed photovoltaic power station cluster multiple times, with a different number of clusters in each clustering;

[0015] Determine the silhouette coefficient of each clustering, and take the cluster with the largest silhouette coefficient as the optimal clustering;

[0016] A reference photovoltaic power station of the photovoltaic power station to be predicted is selected from the cluster where the photovoltaic power station to be predicted is located in the optimal cluster.

[0017] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the step of selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station comprises:

[0018] Determine the distance between the historical power of the photovoltaic power station to be predicted and the historical power of other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located;

[0019] The other photovoltaic power station corresponding to the smallest distance is used as a reference photovoltaic power station of the photovoltaic power station to be predicted.

[0020] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the historical power of the photovoltaic power station to be predicted is the historical power of the photovoltaic power station to be predicted on the second preset date in the season of the preset time period;

[0021] The historical power of the other photovoltaic power stations is the historical power of the other photovoltaic power stations on the second preset date in the season where the preset time period belongs.

[0022] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, before obtaining the first predicted power generation of the reference photovoltaic power station in the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period, the method further includes:

[0023] Determine the correlation between the historical value of each photovoltaic power station in the distributed photovoltaic power station cluster for each meteorological characteristic and the historical power of each photovoltaic power station;

[0024] Determining an average value of all correlations corresponding to each of the meteorological characteristics;

[0025] The meteorological characteristics whose average value is greater than a preset threshold are regarded as key meteorological characteristics affecting the output of the photovoltaic power station.

[0026] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the first predicted power generation of the reference photovoltaic power station in the preset time period is obtained according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period, including:

[0027] Inputting the values ​​of the key meteorological characteristics of the reference photovoltaic power station in a preset time period into a first power generation prediction model, and obtaining a third predicted power generation of the reference photovoltaic power station in the preset time period output by the first power generation prediction model;

[0028] Inputting the value of the key meteorological characteristic of the reference photovoltaic power station within a preset time period into a first error correction model to obtain an error correction value of the third predicted power generation output by the first error correction model;

[0029] The third predicted power generation and the error correction value of the third predicted power generation are added together to obtain the first predicted power generation.

[0030] According to a distributed photovoltaic power station power generation prediction method provided by the present invention, the second predicted power generation of the photovoltaic power station to be predicted within the preset time period is obtained according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation of the reference photovoltaic power station within the preset time period, comprising:

[0031] Inputting the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation into a second power generation prediction model, and obtaining a fourth predicted power generation of the photovoltaic power station to be predicted within the preset time period output by the second power generation prediction model;

[0032] Inputting the value of the key meteorological characteristic of the photovoltaic power station to be predicted within a preset time period and the input first predicted power generation into a second error correction model to obtain an error correction value of the fourth predicted power generation output by the second error correction model;

[0033] The fourth predicted power generation and the error correction value of the fourth predicted power generation are added together to obtain the second predicted power generation.

[0034] The present invention also provides a distributed photovoltaic power station power generation prediction device, comprising:

[0035] A clustering module, used for clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station;

[0036] A first prediction module, configured to obtain a first predicted power generation of the reference photovoltaic power station within a preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station within the preset time period;

[0037] The second prediction module is used to obtain the second predicted power generation power of the photovoltaic power station to be predicted within the preset time period according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation power of the reference photovoltaic power station within the preset time period.

[0038] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a distributed photovoltaic power station power generation prediction method as described above is implemented.

[0039] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the distributed photovoltaic power station power generation prediction method as described in any one of the above is implemented.

[0040] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the distributed photovoltaic power station power generation prediction method as described in any one of the above is implemented.

[0041] The distributed photovoltaic power station power generation prediction method and device provided by the present invention utilizes the fact that most photovoltaic power stations in a distributed photovoltaic power station cluster are in the same or similar external conditions and there is a strong temporal and spatial correlation between the outputs of the photovoltaic power stations. The values ​​of key meteorological characteristics of the photovoltaic power station to be predicted within a preset time period are used as the time characteristics of the photovoltaic power station to be predicted, and the predicted power generation of the reference photovoltaic power station to be predicted within the preset time period is used as the spatial characteristics of the photovoltaic power station to be predicted. The time characteristics and spatial characteristics are comprehensively used to obtain the predicted power generation of the photovoltaic power station to be predicted, thereby improving the prediction accuracy of the power generation of the distributed photovoltaic power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 This is one of the flow charts of the method for predicting the power generation of a distributed photovoltaic power station provided by the present invention;

[0044] Figure 2 It is a schematic diagram of the optimal clustering result in the distributed photovoltaic power station power generation prediction method provided by the present invention;

[0045] Figure 3 It is a structural schematic diagram of the Attention-LSTM model in the distributed photovoltaic power station power generation prediction method provided by the present invention;

[0046] Figure 4 This is the second flow chart of the method for predicting the power generation of a distributed photovoltaic power station provided by the present invention;

[0047] Figure 5 This is the third flow chart of the method for predicting the power generation of a distributed photovoltaic power station provided by the present invention;

[0048] Figure 6 It is a comparative analysis diagram of different input characteristic results on sunny days in the distributed photovoltaic power station power generation prediction method provided by the present invention;

[0049] Figure 7 It is a comparative analysis diagram of different input characteristics of multi-cloud in the distributed photovoltaic power station power generation prediction method provided by the present invention;

[0050] Figure 8 It is a comparative analysis diagram of different input characteristic results on rainy days in the distributed photovoltaic power station power generation prediction method provided by the present invention;

[0051] Fig. 9 It is a structural schematic diagram of a distributed photovoltaic power station power generation prediction device provided by the present invention;

[0052] Fig.10 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Combine the following Figure 1 A method for predicting power generation of a distributed photovoltaic power station according to the present invention is described, comprising:

[0055] Step 101, clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station;

[0056] The distributed photovoltaic power station cluster includes multiple photovoltaic power stations, and the photovoltaic power stations can be clustered according to the historical power of the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain multiple clusters. The clustering method can adopt the K-means algorithm, and this embodiment does not limit the clustering method.

[0057] The photovoltaic power station to be predicted is a photovoltaic power station for which power generation prediction is required. The photovoltaic power station to be predicted is located in a distributed photovoltaic power station cluster.

[0058] The characteristics of the photovoltaic power plants in each cluster are similar, so the reference photovoltaic power plant selected from the cluster where the photovoltaic power plant to be predicted is located has similar characteristics to the photovoltaic power plant to be predicted. Other photovoltaic power plants in the cluster where the photovoltaic power plant to be predicted is located and whose characteristics are closest to the characteristics of the photovoltaic power plant to be predicted can be used as reference photovoltaic power plants.

[0059] Step 102, obtaining a first predicted power generation of the reference photovoltaic power station in the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period;

[0060] The preset time period is a preset future time period. If it is necessary to predict the power generation of the photovoltaic power station to be predicted in the preset time period, the predicted power generation of the reference photovoltaic power station in the preset time period is first predicted, that is, the first predicted power generation.

[0061] The values ​​of the key meteorological characteristics of the reference photovoltaic power station in the preset time period can be obtained in advance through weather forecasts. The first predicted power generation power of the reference photovoltaic power station in the preset time period can be obtained based on the association between the predetermined key meteorological characteristics of the reference photovoltaic power station and the first power generation power of the reference photovoltaic power station, and the values ​​of the key meteorological characteristics of the reference photovoltaic power station in the preset time period.

[0062] Step 103, obtaining a second predicted power generation of the photovoltaic power station to be predicted within the preset time period according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation of the reference photovoltaic power station within the preset time period.

[0063] The values ​​of key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period can be obtained in advance through weather forecast. The second predicted power generation power of the photovoltaic power station to be predicted within the preset time period can be determined based on the predetermined key meteorological characteristic values ​​of the photovoltaic power station to be predicted, the correlation between the first power generation power of the reference photovoltaic power station and the second predicted power generation power of the photovoltaic power station to be predicted, and the key meteorological characteristic values ​​of the photovoltaic power station to be predicted within the preset time period and the first power generation power of the reference photovoltaic power station within the preset time period.

[0064] This embodiment utilizes the fact that most of the photovoltaic power stations in a distributed photovoltaic power station cluster are in the same or similar external conditions and there is a strong temporal and spatial correlation between the outputs of the photovoltaic power stations. The values ​​of key meteorological characteristics of the photovoltaic power station to be predicted within a preset time period are used as the time characteristics of the photovoltaic power station to be predicted, and the predicted power generation power of the reference photovoltaic power station to be predicted within the preset time period is used as the spatial characteristics of the photovoltaic power station to be predicted. The time characteristics and spatial characteristics are comprehensively used to obtain the predicted power generation power of the photovoltaic power station to be predicted, thereby improving the prediction accuracy of the power generation power of the distributed photovoltaic power station.

[0065] Based on the above embodiment, the photovoltaic power stations in the distributed photovoltaic power station cluster are clustered to obtain multiple clusters in this embodiment, including:

[0066] Selecting the historical power of the first preset date of each season from the annual historical power of the photovoltaic power station;

[0067] Determine one or more of the mean, standard deviation, coefficient of variation, kurtosis and skewness of the daily historical power in the historical power of each first preset date;

[0068] The photovoltaic power station is clustered to obtain a plurality of clusters according to one or more of the mean value, standard deviation, coefficient of variation, kurtosis and skewness.

[0069] The annual historical power of each photovoltaic power station in the distributed photovoltaic power station cluster can be standardized, and the historical power of N typical days in each season is selected from the standardized annual historical power.

[0070] The normalized historical power for the whole year can be subjected to curve fitting, from which historical power curves of N typical days can be intercepted.

[0071] For example, May 15 (spring), August 15 (summer), November 15 (fall), and February 15 (winter) are selected as typical days.

[0072] Calculate one or more of the average, standard deviation, coefficient of variation, kurtosis and skewness of the daily historical power. The definition formulas of the coefficient of variation C, kurtosis K and skewness S are as follows:

[0073]

[0074] Where σ is the standard deviation of daily historical power, is the average value of daily historical power, n is the total number of historical power collected daily, X i is the i-th historical power collected daily.

[0075] One or more of the average value, standard deviation, coefficient of variation, kurtosis and skewness of the 4N-day historical power selected for each photovoltaic power station is used as clustering features to cluster the photovoltaic power stations in the distributed photovoltaic power station cluster.

[0076] On the basis of the above embodiment, in this embodiment, the photovoltaic power stations in the distributed photovoltaic power station cluster are clustered to obtain multiple clusters, and a reference photovoltaic power station of the photovoltaic power station to be predicted is selected from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station, including:

[0077] Clustering the photovoltaic power stations in the distributed photovoltaic power station cluster multiple times, with a different number of clusters in each clustering;

[0078] Determine the silhouette coefficient of each clustering, and take the cluster with the largest silhouette coefficient as the optimal clustering;

[0079] A reference photovoltaic power station of the photovoltaic power station to be predicted is selected from the cluster where the photovoltaic power station to be predicted is located in the optimal cluster.

[0080] The photovoltaic power stations in the distributed photovoltaic power station cluster are clustered multiple times with different numbers of clusters, and the silhouette coefficient of each cluster is analyzed. The cluster with the largest silhouette coefficient is taken as the optimal cluster, and the clustering result of the optimal cluster is taken as the division result of the photovoltaic power stations in the distributed photovoltaic power station cluster. The calculation formula of the silhouette coefficient L is as follows:

[0081]

[0082] Among them, a represents the average distance from the clustering features of the photovoltaic power station in each cluster to the clustering features of other photovoltaic power stations in the same cluster, that is, the intra-cluster distance; b represents the minimum value of the average distance from the clustering features of the photovoltaic power station in each cluster to the clustering features of photovoltaic power stations in other clusters, that is, the inter-cluster distance; max(a,b) is the larger value of a and b. The closer the value of the silhouette coefficient is to 1, the better the clustering effect is, and the closer it is to -1, the worse the clustering effect is.

[0083] For example, the K-means algorithm is used to cluster the photovoltaic power stations in the distributed photovoltaic power station cluster three times. The corresponding cluster numbers are 2, 3 and 4, and the corresponding silhouette coefficients are 0.54, 0.83 and 0.64, respectively. The three clusters obtained by clustering using the K-means algorithm are taken as the optimal clustering results. The schematic diagram of the optimal clustering result is shown in the figure below. Figure 2 The reference photovoltaic power station of the photovoltaic power station to be predicted is selected based on the optimal clustering result.

[0084] On the basis of the above embodiment, the step of selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted in the photovoltaic power station is located in this embodiment includes:

[0085] Determine the distance between the historical power of the photovoltaic power station to be predicted and the historical power of other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located;

[0086] The other photovoltaic power station corresponding to the smallest distance is used as a reference photovoltaic power station of the photovoltaic power station to be predicted.

[0087] Calculate the distance between the historical power of the photovoltaic power station to be predicted and other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located at multiple same times, such as the Euclidean distance. The calculation formula of the Euclidean distance dist is as follows:

[0088]

[0089] Among them, X is the historical power sequence of the photovoltaic power station to be predicted, and Y is the historical power sequence of other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located. The lengths of X and Y are the same, m2 is the length of X and Y, and x i is the i-th historical power in X, y i is the i-th historical power in Y.

[0090] The other photovoltaic power stations with the smallest Euclidean distance to the photovoltaic power station to be predicted are selected as reference photovoltaic power stations.

[0091] On the basis of the above embodiment, the historical power of the photovoltaic power station to be predicted in this embodiment is the historical power of the photovoltaic power station to be predicted on the second preset date in the season of the preset time period;

[0092] The historical power of the other photovoltaic power stations is the historical power of the other photovoltaic power stations on the second preset date in the season where the preset time period belongs.

[0093] Determine the season in which the future preset time period for predicting the power generation of the photovoltaic power station to be predicted is located. Select M-day historical power from the historical power of the season, such as selecting the historical power of a typical day. Calculate the distance between the M-day historical power of the photovoltaic power station to be predicted and other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located.

[0094] For example, May 15 (spring), August 15 (summer), November 15 (autumn), and February 15 (winter) are selected as typical days. If the season in which the preset time period is located is not summer, the distance between the historical power of the photovoltaic power station to be predicted and other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located on August 15 is calculated. The other photovoltaic power station with the smallest distance to the photovoltaic power station to be predicted is selected as the reference photovoltaic power station.

[0095] Assuming that PV station No. 1 is used as the PV station to be predicted, the sum of the Euclidean distances between the PV station to be predicted and the typical daily historical power of other PV stations in the cluster where the PV station to be predicted is shown in Table 1. From Table 1, it can be seen that the Euclidean distance between PV station No. 10 and the typical daily historical power of the PV station to be predicted is the smallest, so PV station No. 10 is selected as the reference PV station for PV station No. 1.

[0096] Table 1 Example of Euclidean distance between photovoltaic power stations

[0097]

[0098]

[0099] On the basis of the above embodiments, in this embodiment, before obtaining the first predicted power generation power of the reference photovoltaic power station in the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period, the method further includes:

[0100] Determine the correlation between the historical value of each photovoltaic power station in the distributed photovoltaic power station cluster for each meteorological characteristic and the historical power of each photovoltaic power station;

[0101] Determining an average value of all correlations corresponding to each of the meteorological characteristics;

[0102] The meteorological characteristics whose average value is greater than a preset threshold are regarded as key meteorological characteristics affecting the output of the photovoltaic power station.

[0103] The historical values ​​of various meteorological characteristics of each photovoltaic power station in the distributed photovoltaic power station cluster throughout the year can be standardized. The correlation between the historical values ​​of various meteorological characteristics of each photovoltaic power station after standardization and the historical power of each photovoltaic power station, such as the Pearson coefficient, is calculated. The calculation formula of the Pearson coefficient is as follows:

[0104]

[0105] in, Represents the covariance of the historical value sequence X and the historical power sequence Y of a certain meteorological characteristic of a photovoltaic power station, and They represent the standard deviation of X and Y respectively, m1 represents the length of X and Y, and X i represents the i-th historical value of a certain meteorological feature in X, Y i represents the i-th historical power in Y, For all X i The average value of For all Y i The closer the absolute value of the Pearson coefficient is to 1, the stronger the correlation between the variables is, and the closer it is to 0, the weaker the correlation between the variables is.

[0106] The average Pearson coefficient of all photovoltaic power stations for each meteorological feature is calculated, and the meteorological features with an average Pearson coefficient greater than a preset threshold, such as 0.6, are regarded as key meteorological features.

[0107] Taking the photovoltaic power station in photovoltaic cluster No. 1 as an example, the calculation results of the Pearson coefficient are shown in Table 2. It can be seen that the global horizontal irradiance (Global Horizontal Irradiance, GHI) has a strong correlation with photovoltaic power, the direct normal irradiance (Direct Normal Irradiance, DNI) and the solar zenith angle have a strong correlation with photovoltaic power, the relative humidity has a moderate correlation with photovoltaic power, the direct horizontal irradiance (Direct Horizontal Irradiance, DHI), wind speed, dew point, temperature and photovoltaic power have a weak correlation, and the pressure, wind direction, surface albedo and photovoltaic power have almost no correlation. Therefore, four features with high correlation, namely GHI, DNI, solar zenith angle, and relative humidity are selected as the main meteorological output parameters affecting photovoltaic power generation.

[0108] Table 2 Examples of Pearson coefficients for various meteorological characteristics

[0109] Weather characteristics Pearson coefficient Weather characteristics Pearson coefficient DHI 0.192 temperature 0.311 DNI 0.683 pressure -0.006 Dew point -0.205 GHI 0.851 Surface albedo 0.128 wind direction 0.062 Phoenix Speed 0.255 Solar zenith angle -0.693 Relative humidity -0.476

[0110] On the basis of the above embodiments, the method of obtaining the first predicted power generation of the reference photovoltaic power station in the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period in this embodiment includes:

[0111] Inputting the values ​​of the key meteorological characteristics of the reference photovoltaic power station in a preset time period into a first power generation prediction model, and obtaining a third predicted power generation of the reference photovoltaic power station in the preset time period output by the first power generation prediction model;

[0112] Inputting the value of the key meteorological characteristic of the reference photovoltaic power station within a preset time period into a first error correction model to obtain an error correction value of the third predicted power generation output by the first error correction model;

[0113] The third predicted power generation and the error correction value of the third predicted power generation are added together to obtain the first predicted power generation.

[0114] The first power generation prediction model and the first error correction model can both be Attention-LSTM (Attention Long Short Term Memory) models. The structure of the Attention-LSTM model is as follows: Figure 3 As shown, it includes two LSTM layers, two Dropout layers, one Attention layer and one Dense (fully connected) layer.

[0115] The key meteorological characteristic values ​​of the photovoltaic power station samples within a period of time are used as samples, and the actual power generation of the photovoltaic power station samples within a period of time is used as a label to train the first power generation prediction model. Specifically, the key meteorological characteristic values ​​of the photovoltaic power station samples within a period of time are input into the first power generation prediction model to obtain the predicted power generation of the photovoltaic power station samples within a period of time output by the first power generation prediction model. According to the difference between the predicted power generation and the actual power generation of the photovoltaic power station samples within a period of time, the parameters of the first power generation prediction model are adjusted so that the difference between the predicted power generation and the actual power generation gradually decreases.

[0116] The key meteorological characteristic values ​​of the photovoltaic power station samples within a period of time are input into the trained first power generation prediction model to obtain the predicted power generation of the photovoltaic power station samples within a period of time output by the trained first power generation prediction model. The difference between the predicted power generation and the actual power generation is used as a label, and the key meteorological characteristic values ​​of the photovoltaic power station samples within a period of time are used as samples to train the first error correction model.

[0117] The key meteorological characteristic values ​​of the photovoltaic power station samples over a period of time are input into the first error correction model to obtain the error correction value output by the first error correction model. According to the distance between the error correction value and the difference label, the parameters of the first error correction model are adjusted so that the error correction value gradually approaches the difference label.

[0118] The values ​​of the key meteorological characteristics of the reference photovoltaic power station in the preset time period are respectively input into the trained first power generation prediction model and the trained first error correction model to obtain the predicted power generation and error correction value of the reference photovoltaic power station in the preset time period. The two are added together to obtain the final predicted power generation of the reference photovoltaic power station, that is, the first predicted power generation.

[0119] On the basis of the above embodiments, the method of obtaining the second predicted power generation power of the photovoltaic power station to be predicted within the preset time period according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation power of the reference photovoltaic power station within the preset time period in this embodiment includes:

[0120] Inputting the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation into a second power generation prediction model, and obtaining a fourth predicted power generation of the photovoltaic power station to be predicted within the preset time period output by the second power generation prediction model;

[0121] Inputting the value of the key meteorological characteristic of the photovoltaic power station to be predicted within a preset time period and the input first predicted power generation into a second error correction model to obtain an error correction value of the fourth predicted power generation output by the second error correction model;

[0122] The fourth predicted power generation and the error correction value of the fourth predicted power generation are added together to obtain the second predicted power generation.

[0123] The second power generation prediction model and the second error correction model can both be Attention-LSTM (Attention Long Short Term Memory) models. The structure of the Attention-LSTM model is as follows: Figure 3 The process of using the second power generation prediction model and the second error correction model to predict power generation is as follows: Figure 4 shown.

[0124] like Figure 5As shown, the key meteorological characteristic values ​​of the photovoltaic power station samples in group A over a period of time and the predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples are used as samples, and the actual power generation of the photovoltaic power station samples over a period of time is used as a label to train the second power generation prediction model. The predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples is predicted using the trained first power generation prediction model.

[0125] Specifically, the key meteorological characteristic values ​​of the photovoltaic power station samples within a period of time and the predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples are input into the second power generation prediction model to obtain the predicted power generation of the photovoltaic power station samples within a period of time output by the second power generation prediction model. According to the difference between the predicted power generation of the photovoltaic power station samples within a period of time and the actual power generation, the parameters of the second power generation prediction model are adjusted so that the difference between the predicted power generation and the actual power generation gradually decreases.

[0126] The key meteorological characteristic values ​​of the photovoltaic power station samples in group B over a period of time and the predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples are input into the trained second power generation prediction model to obtain the predicted power generation of the photovoltaic power station samples over a period of time output by the trained second power generation prediction model. The difference between the predicted power generation and the actual power generation is used as a label, and the key meteorological characteristic values ​​of the photovoltaic power station samples over a period of time are used as samples to train the second error correction model.

[0127] Specifically, the key meteorological characteristic values ​​of the photovoltaic power station samples in group B over a period of time and the predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples are input into the second error correction model to obtain the error correction value output by the second error correction model. According to the distance between the error correction value and the difference label, the parameters of the second error correction model are adjusted so that the error correction value gradually approaches the difference label.

[0128] The key meteorological characteristic values ​​of the photovoltaic power station samples in group C over a period of time and the predicted power generation of the reference photovoltaic power station of the photovoltaic power station samples are input into the trained second power generation prediction model and the trained second error correction model to obtain the predicted power generation and error correction value of the photovoltaic power station samples in group C. The two are added together to obtain the final predicted power generation of the photovoltaic power station in group C. The second power generation prediction model and the second error correction model are tested according to the final predicted power generation of the photovoltaic power station in group C.

[0129] The values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation of the reference photovoltaic power station are input into the trained second power generation prediction model and the trained second error correction model to obtain the predicted power generation and error correction value of the photovoltaic power station to be predicted in the preset time period. The two are added to obtain the final predicted power generation of the photovoltaic power station to be predicted, that is, the second predicted power generation.

[0130] In order to verify the effectiveness of using spatiotemporal fusion features as the input of the prediction model, the prediction model using spatiotemporal fusion features as input is compared and analyzed with the prediction model using only meteorological features or power features as input.

[0131] Taking the No. 1 power station in the No. 1 photovoltaic cluster as an example, the experimental results are as follows: Figure 6 , Figure 7 , Figure 8 And as shown in Table 3. Figure 6 , Figure 7 and Figure 8 The photovoltaic power prediction curves from 7:00 to 19:00 on November 17 (sunny), November 12 (cloudy) and November 28 (rainy) are selected for comparison and display. Table 3 shows the error comparison under different input characteristics.

[0132] Table 3 Comparison of prediction model errors with different input features

[0133] Input Features MAE / MW MSE / MW Spatiotemporal fusion features 2.71 16.72 Meteorological characteristics 2.93 18.91 Power characteristics 3.26 21.33

[0134] As shown in Table 3, the MAE (Mean Absolute Error) and MSE (Mean Squared Error) of the method provided in this embodiment using the spatiotemporal information fusion features as the input of the prediction model are 2.71MW and 16.72MW respectively. Compared with meteorological features, MAE and MSE are reduced by 7.51% and 11.58% respectively; compared with power features, MAE and MSE are reduced by 16.87% and 21.61% respectively. It can be seen that using spatiotemporal information fusion features as the input of the prediction model can effectively improve the accuracy of distributed photovoltaic power station power generation prediction.

[0135] The distributed photovoltaic power station power generation prediction device provided by the present invention is described below. The distributed photovoltaic power station power generation prediction device described below and the distributed photovoltaic power station power generation prediction method described above can be referenced to each other.

[0136] like Fig. 9 As shown, the distributed photovoltaic power station power generation prediction device includes a clustering module 901, a first prediction module 902 and a second prediction module 903, wherein:

[0137] The clustering module 901 is used to cluster the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain multiple clusters, and select a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station;

[0138] The first prediction module 902 is used to obtain a first predicted power generation of the reference photovoltaic power station within the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station within the preset time period;

[0139] The second prediction module 903 is used to obtain the second predicted power generation of the photovoltaic power station to be predicted within the preset time period based on the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation of the reference photovoltaic power station within the preset time period.

[0140] This embodiment utilizes the fact that most of the photovoltaic power stations in a distributed photovoltaic power station cluster are in the same or similar external conditions and there is a strong temporal and spatial correlation between the outputs of the photovoltaic power stations. The values ​​of key meteorological characteristics of the photovoltaic power station to be predicted within a preset time period are used as the time characteristics of the photovoltaic power station to be predicted, and the predicted power generation power of the reference photovoltaic power station to be predicted within the preset time period is used as the spatial characteristics of the photovoltaic power station to be predicted. The time characteristics and spatial characteristics are comprehensively used to obtain the predicted power generation power of the photovoltaic power station to be predicted, thereby improving the prediction accuracy of the power generation power of the distributed photovoltaic power station.

[0141] Fig.10 An example of a physical structure diagram of an electronic device is shown in FIG. Fig.10As shown, the electronic device may include: a processor 1010, a communication interface 1020, a memory 1030 and a communication bus 1040, wherein the processor 1010, the communication interface 1020 and the memory 1030 communicate with each other through the communication bus 1040. The processor 1010 may call the logic instructions in the memory 1030 to execute the distributed photovoltaic power station power generation prediction method, the method comprising: clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain multiple clusters, selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station; according to the correlation between the historical values ​​of various meteorological characteristics of the photovoltaic power station and the historical power of the photovoltaic power station, determining whether the various meteorological characteristics are key meteorological characteristics that affect the output of the photovoltaic power station; according to the value of the key meteorological characteristics of the reference photovoltaic power station in the preset time period, obtaining the first predicted power generation of the reference photovoltaic power station in the preset time period; according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation of the reference photovoltaic power station in the preset time period, obtaining the second predicted power generation of the photovoltaic power station to be predicted in the preset time period.

[0142] In addition, the logic instructions in the above-mentioned memory 1030 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0143] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the distributed photovoltaic power station power generation prediction method provided by the above methods, the method including: clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain multiple clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted in the photovoltaic power station is located; according to the correlation between the historical values ​​of various meteorological characteristics of the photovoltaic power station and the historical power of the photovoltaic power station, determining whether the various meteorological characteristics are key meteorological characteristics that affect the output of the photovoltaic power station; according to the values ​​of the key meteorological characteristics of the reference photovoltaic power station in the preset time period, obtaining the first predicted power generation of the reference photovoltaic power station in the preset time period; according to the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation of the reference photovoltaic power station in the preset time period, obtaining the second predicted power generation of the photovoltaic power station to be predicted in the preset time period.

[0144] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the distributed photovoltaic power station power generation prediction method provided by the above-mentioned methods, the method comprising: clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain multiple clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted in the photovoltaic power station is located; determining whether the various meteorological characteristics are key meteorological characteristics that affect the output of the photovoltaic power station based on the correlation between the historical values ​​of various meteorological characteristics of the photovoltaic power station and the historical power of the photovoltaic power station; obtaining a first predicted power generation power of the reference photovoltaic power station within the preset time period based on the values ​​of the key meteorological characteristics of the reference photovoltaic power station within the preset time period; obtaining a second predicted power generation power of the photovoltaic power station to be predicted within the preset time period based on the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation power of the reference photovoltaic power station within the preset time period.

[0145] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0146] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting power generation of a distributed photovoltaic power station, characterized in that: include: Clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station; Obtaining a first predicted power generation of the reference photovoltaic power station within the preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station within the preset time period; According to the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted in the preset time period and the first predicted power generation of the reference photovoltaic power station in the preset time period, the second predicted power generation of the photovoltaic power station to be predicted in the preset time period is obtained.

2. The method for predicting power generation of a distributed photovoltaic power station according to claim 1, characterized in that: The photovoltaic power stations in the distributed photovoltaic power station cluster are clustered to obtain multiple clusters, including: Selecting the historical power of the first preset date of each season from the annual historical power of the photovoltaic power station; determining one or more of a mean, a standard deviation, a coefficient of variation, a kurtosis, and a skewness of the historical power for each first predetermined date; The photovoltaic power station is clustered to obtain a plurality of clusters according to one or more of the mean value, standard deviation, coefficient of variation, kurtosis and skewness.

3. The method for predicting power generation of a distributed photovoltaic power station according to claim 1, characterized in that: The step of clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located, comprises: Clustering the photovoltaic power stations in the distributed photovoltaic power station cluster multiple times, with a different number of clusters in each clustering; Determine the silhouette coefficient of each clustering, and take the cluster with the largest silhouette coefficient as the optimal clustering; A reference photovoltaic power station of the photovoltaic power station to be predicted is selected from the cluster where the photovoltaic power station to be predicted is located in the optimal cluster.

4. The method for predicting power generation of a distributed photovoltaic power station according to claim 1, characterized in that: The step of selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station comprises: Determine the distance between the historical power of the photovoltaic power station to be predicted and the historical power of other photovoltaic power stations in the cluster where the photovoltaic power station to be predicted is located; The other photovoltaic power station corresponding to the smallest distance is used as a reference photovoltaic power station of the photovoltaic power station to be predicted.

5. The method for predicting power generation of a distributed photovoltaic power station according to claim 4, characterized in that: The historical power of the photovoltaic power station to be predicted is the historical power of the photovoltaic power station to be predicted on the second preset date in the season of the preset time period; The historical power of the other photovoltaic power stations is the historical power of the other photovoltaic power stations on the second preset date in the season where the preset time period belongs.

6. The method for predicting power generation of a distributed photovoltaic power station according to any one of claims 1 to 5, characterized in that: Before obtaining the first predicted power generation of the reference photovoltaic power station in the preset time period according to the value of the key meteorological characteristic of the reference photovoltaic power station in the preset time period, the method further includes: Determine the correlation between the historical value of each photovoltaic power station in the distributed photovoltaic power station cluster for each meteorological characteristic and the historical power of each photovoltaic power station; Determining an average value of all correlations corresponding to each of the meteorological characteristics; The meteorological characteristics whose average value is greater than a preset threshold are regarded as key meteorological characteristics affecting the output of the photovoltaic power station.

7. The method for predicting power generation of a distributed photovoltaic power station according to any one of claims 1 to 5, characterized in that: The step of obtaining a first predicted power generation power of the reference photovoltaic power station within a preset time period according to a value of a key meteorological characteristic of the reference photovoltaic power station within the preset time period comprises: Inputting the values ​​of the key meteorological characteristics of the reference photovoltaic power station in a preset time period into a first power generation prediction model, and obtaining a third predicted power generation of the reference photovoltaic power station in the preset time period output by the first power generation prediction model; Inputting the value of the key meteorological characteristic of the reference photovoltaic power station within a preset time period into a first error correction model to obtain an error correction value of the third predicted power generation output by the first error correction model; The third predicted power generation and the error correction value of the third predicted power generation are added together to obtain the first predicted power generation.

8. The method for predicting power generation of a distributed photovoltaic power station according to any one of claims 1 to 5, characterized in that: The step of obtaining a second predicted power generation power of the photovoltaic power station to be predicted within the preset time period according to the value of the key meteorological characteristic of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation power of the reference photovoltaic power station within the preset time period comprises: Inputting the values ​​of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation into a second power generation prediction model, and obtaining a fourth predicted power generation of the photovoltaic power station to be predicted within the preset time period output by the second power generation prediction model; Inputting the value of the key meteorological characteristic of the photovoltaic power station to be predicted within a preset time period and the input first predicted power generation into a second error correction model to obtain an error correction value of the fourth predicted power generation output by the second error correction model; The fourth predicted power generation and the error correction value of the fourth predicted power generation are added together to obtain the second predicted power generation.

9. A distributed photovoltaic power station power generation prediction device, characterized in that: include: A clustering module, used for clustering the photovoltaic power stations in the distributed photovoltaic power station cluster to obtain a plurality of clusters, and selecting a reference photovoltaic power station of the photovoltaic power station to be predicted from the cluster where the photovoltaic power station to be predicted is located in the photovoltaic power station; A first prediction module, configured to obtain a first predicted power generation of the reference photovoltaic power station within a preset time period according to the value of the key meteorological characteristics of the reference photovoltaic power station within the preset time period; The second prediction module is used to obtain the second predicted power generation power of the photovoltaic power station to be predicted within the preset time period according to the value of the key meteorological characteristics of the photovoltaic power station to be predicted within the preset time period and the first predicted power generation power of the reference photovoltaic power station within the preset time period.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for predicting the power generation of a distributed photovoltaic power station as claimed in any one of claims 1 to 8 is implemented.

Images