Renewable energy power generation power prediction method, device and system, and storage medium

Abstract

This application provides a method, apparatus, system, and storage medium for predicting renewable energy power generation, belonging to the field of renewable energy power generation technology. The method includes acquiring a multi-source dataset, which includes power generation data and spatial feature data of each photovoltaic (PV) module, as well as regional meteorological data of the area where the PV power generation system is located; filtering meteorological influencing factors to obtain target meteorological influencing factors; constructing a graph structure with the geographical location of each PV module as nodes and the temporal and spatial correlations between the spatial feature data and power generation data of each PV module as edges; adjusting the temporal and spatial correlations in the graph structure using a graph neural network, and then fusing the multi-source dataset to obtain a fused feature set; and obtaining the power generation prediction result based on the fused feature set and a PV power prediction model. This application can improve the accuracy of PV power generation prediction.

Description

Technical Field

[0001] This application belongs to the field of renewable energy power generation technology, and more specifically, relates to a renewable energy power generation prediction method, device, system, and storage medium. Background Technology

[0002] Against the backdrop of today's global energy transition, renewable energy power generation, with its significant advantages of being clean and sustainable, is gradually becoming a key development direction in the energy sector, with its development momentum becoming increasingly strong and its share in the energy structure continuing to rise. Among them, photovoltaic power generation, as an important component of renewable energy power generation, has been widely used in many fields due to its convenience and abundant resources.

[0003] Existing photovoltaic power generation prediction methods typically rely on prediction models. Many traditional prediction models often only consider a few key factors, such as making predictions based solely on historical power generation data and simple solar irradiance data, ignoring the combined effects of multiple factors such as temperature, weather changes, and equipment status. This leads to a significant deviation between the prediction results and the actual power generation, affecting the accuracy of photovoltaic power generation prediction and hindering its widespread application. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, system, and storage medium for predicting renewable energy power generation, in order to solve one or more technical problems mentioned in the background art, thereby improving the accuracy of photovoltaic power generation prediction.

[0005] A first aspect of this application provides a renewable energy power generation prediction method, applied to a processor in a photovoltaic power generation system, the photovoltaic power generation system further including multiple photovoltaic modules; the renewable energy power generation prediction method includes: Obtain multi-source datasets, which include power generation data and spatial feature data of each photovoltaic module within the first historical time period, as well as regional meteorological data of the area where the photovoltaic power generation system is located; the spatial feature data represents the location and status characteristics of the photovoltaic modules in the spatial dimension; Meteorological influencing factors are screened based on the spatiotemporal correlation between the power generation data of each photovoltaic module and regional meteorological data to obtain target meteorological influencing factors; A graph structure is constructed using the geographical location of each photovoltaic module as nodes and the temporal and spatial relationships between the spatial characteristic data and the power generation data of each photovoltaic module as edges. Using the target meteorological influencing factors as global features, a graph neural network is used to adjust the temporal and spatial relationships in the graph structure. Based on the adjusted temporal and spatial relationships, the multi-source datasets are fused to obtain a fused feature set. The power generation prediction results for the second time period in the future are obtained based on the fused feature set and the photovoltaic power prediction model.

[0006] A second aspect of this application provides a renewable energy power generation prediction device, applied to a processor in a photovoltaic power generation system, the photovoltaic power generation system further including multiple photovoltaic modules; the renewable energy power generation prediction device includes: The data acquisition module is used to acquire multi-source datasets, which include power generation data and spatial feature data of each photovoltaic module within the first historical time period, as well as regional meteorological data of the area where the photovoltaic power generation system is located; the spatial feature data represents the location and status characteristics of the photovoltaic modules in the spatial dimension. The data filtering module is used to filter meteorological influencing factors based on the spatiotemporal correlation between the power generation data of each photovoltaic module and regional meteorological data, so as to obtain the target meteorological influencing factors. The graph structure construction module is used to construct a graph structure with the geographical location of each photovoltaic module as nodes and the temporal and spatial relationships between the spatial feature data and the power generation data of each photovoltaic module as edges. The data fusion module is used to take the target meteorological influencing factors as global features, and use graph neural networks to adjust the temporal and spatial relationships in the graph structure. Based on the adjusted temporal and spatial relationships, the multi-source datasets are fused to obtain a fused feature set. The prediction module is used to obtain the power generation prediction results for the second time period in the future based on the fused feature set and the photovoltaic power prediction model.

[0007] A third aspect of this application provides a photovoltaic power generation system, including multiple photovoltaic modules, a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-described renewable energy power generation prediction method.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described renewable energy power generation prediction method.

[0009] The beneficial effects of the renewable energy power generation prediction method, apparatus, system, and storage medium provided in this application are as follows: This application's embodiments filter target meteorological influencing factors through spatiotemporal correlation, eliminating redundant or irrelevant variables and reducing input noise. Simultaneously, this application's embodiments construct a graph structure using the geographical location of photovoltaic modules as nodes and spatiotemporal correlations as edges, and utilize a graph neural network to dynamically adjust the relationships within the graph structure, enabling deep integration of power generation data, spatial feature data, and regional meteorological data. Compared to traditional single-model or simple feature overlay methods, this application's embodiments can effectively capture the spatial dependence between photovoltaic modules and the influence of meteorology, significantly improving the accuracy and stability of power generation prediction.

[0010] In addition, this application embodiment extracts location and state features from spatial feature data and combines them with graph neural networks to adaptively adjust the spatiotemporal correlation, which can adapt to the layout differences of different photovoltaic power plants and the dynamic changes in meteorological conditions, thereby enhancing the model's adaptability to the complex and ever-changing actual operating environment. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic flowchart illustrating a renewable energy power generation prediction method provided in an embodiment of this application; Figure 2 This is a structural block diagram of a renewable energy power generation prediction device provided in an embodiment of this application; Figure 3 This is a schematic block diagram of a photovoltaic power generation system provided in an embodiment of this application. Detailed Implementation

[0013] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0014] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0015] Please refer to Figure 1 , Figure 1This is a flowchart illustrating a renewable energy power generation prediction method provided in an embodiment of this application. The method can be executed by a processor in a photovoltaic power generation system and may include steps S101-S105.

[0016] S101: Obtain multi-source datasets, which include power generation data and spatial feature data of each photovoltaic module within the first historical time period, as well as regional meteorological data of the area where the photovoltaic power generation system is located.

[0017] Among them, spatial feature data represents the location and state characteristics of photovoltaic modules in the spatial dimension.

[0018] In this embodiment, the photovoltaic module is the core unit of the photovoltaic power generation system, capable of directly converting solar energy into electrical energy through photoelectric conversion. The photovoltaic power generation system also includes devices for monitoring the electrical parameters of each photovoltaic module, such as current sensors and voltage sensors. These devices send the collected electrical parameters to a processor, which can then calculate the power generation data of each photovoltaic module, such as instantaneous power generation value, time-series power generation sequence, average power generation value within a first time period, and identification information (timestamp and photovoltaic module number) corresponding to each sampling time.

[0019] Spatial feature data represents the fixed structure and real-time status of photovoltaic modules in space, including two categories: location features and status features. Location features include fixed geometric and layout information such as array layout, tilt angle, azimuth angle, spacing, orientation, shading boundary, and support structure. Status features include spatial information that changes with status, such as module temperature, dust cover thickness, aging coefficient, fault status, and cloud shading status.

[0020] Regional meteorological data for the area where the photovoltaic power generation system is located includes solar irradiance, temperature, humidity, wind speed, wind direction, precipitation, snowfall, air pressure, and haze. These meteorological data are external factors affecting the photovoltaic power generation output. The area where the photovoltaic power generation system is located can be a circular area with radius r, centered on the photovoltaic power station and covering all photovoltaic power generation modules, or an area determined by other regional division methods.

[0021] The power generation data and spatial characteristic data of each photovoltaic module within the aforementioned historical first time period, along with regional meteorological data, are preprocessed through data cleaning, standardization, smoothing, and denoising to obtain a multi-source dataset. In this embodiment, data cleaning may include missing value handling and outlier handling. Missing value handling is mainly achieved through linear interpolation, while outlier handling is mainly achieved through statistical methods such as the 3σ principle or k-means clustering. Data standardization can be achieved through Min-Max standardization or Z-Score standardization. Data smoothing and denoising can be achieved through wavelet thresholding. These methods are mostly conventional in the field, and their specific implementations will not be elaborated upon in this paper.

[0022] S102: Based on the spatiotemporal correlation between the power generation data of each photovoltaic module and the regional meteorological data, meteorological influencing factors are screened to obtain target meteorological influencing factors.

[0023] In this embodiment, the power generation of photovoltaic modules is significantly affected by meteorological conditions, and different meteorological conditions have different impacts on the power generation of photovoltaic modules. Meteorological influencing factors may include solar irradiance, temperature, humidity, wind speed, wind direction, wind force, precipitation, snowfall, air pressure, and haze. For example, solar irradiance and precipitation have a greater impact on the power generation of photovoltaic modules than wind force and wind speed.

[0024] This embodiment can calculate the sum of the power generation of each photovoltaic module to obtain the total power generation. Using the total power generation as the dependent variable and regional meteorological data as the independent variable, the response relationship at different times is established in the time dimension, and the difference in the influence of meteorological field distribution on photovoltaic modules at different locations is established in the spatial dimension, thereby quantifying the spatiotemporal correlation between each meteorological influencing factor and the power generation of the module.

[0025] Spatiotemporal correlation includes temporal correlation and spatial correlation. Both parameters can be calculated using the Pearson correlation coefficient, mutual information, or time-delay cross-correlation coefficient between the independent and dependent variables. Mutual information is used to measure the nonlinear time-series dependence between meteorological influencing factors and power generation; the time-delay cross-correlation coefficient is used to measure the lagged effect of changes in meteorological influencing factors on power generation, and can be obtained by calculating the cross-correlation coefficient between the independent and dependent variables at different time offsets.

[0026] In this embodiment, the time correlation and spatial correlation are weighted and summed to obtain the total correlation between each meteorological influencing factor and the power generation data of each photovoltaic module. The total correlation is then sorted, and the top N meteorological influencing factors with the largest total correlation are selected as the target meteorological influencing factors.

[0027] S103: Construct a graph structure with the geographical location of each photovoltaic module as nodes and the temporal and spatial correlations between the spatial characteristic data and the power generation data of each photovoltaic module as edges.

[0028] In this embodiment, the temporal correlation between the spatial characteristic data and the power generation data of each photovoltaic module can be extracted by time series analysis; the spatial correlation between the spatial characteristic data and the power generation data of each photovoltaic module can be obtained by spatial autocorrelation analysis.

[0029] In this embodiment, the time series analysis method can be cross-correlation function (CCF). For each photovoltaic module, its spatial characteristic data (e.g., module tilt angle and aging coefficient, which change slowly over time or are considered constant) are subjected to lag correlation calculation with the power generation data. The correlation coefficient between the two data series at different lag orders is calculated to obtain the significantly correlated lag time. The output of cross-correlation analysis is an edge with a time delay. For example, if a 12-hour lag correlation is found between the spatial characteristics of a photovoltaic module (e.g., the thickness of dust cover) and the power reduction (because dust accumulation requires a certain amount of sunlight to manifest the loss), then an edge with a time delay is formed. In this embodiment, this time delay is the time correlation relationship between the two photovoltaic modules.

[0030] Time series analysis can also employ Granger causality tests. A vector autoregression model is established to examine whether changes in spatial feature data contribute to the prediction of changes in power generation data. If historical values ​​of spatial features significantly improve the accuracy of power generation prediction, a temporal causal relationship between spatial features and power generation is considered to exist. The output then shows a directed time correlation, along with the optimal lag order (first-order lag can be 12 hours, second-order lag is 24 hours, and so on). This optimal lag order represents the temporal correlation between two photovoltaic modules.

[0031] Spatial autocorrelation analysis can be performed using global and local Moran's index methods. The specific analysis process is as follows: Using the geographical location (latitude and longitude or array row and column coordinates) of each photovoltaic module as the spatial unit, calculate the Euclidean distance between each pair of photovoltaic modules. Use the reciprocal of this Euclidean distance as the spatial weight to construct a spatial weight matrix. Calculate the global Moran's index of power generation and determine whether overall spatial autocorrelation exists based on statistical results. Calculate the local Moran's index of power generation to identify spatial clustering or anomalous patterns between each photovoltaic module and its neighboring photovoltaic modules. For significantly correlated photovoltaic module pairs, establish spatial edges, with edge weights being the spatial autocorrelation coefficients. This spatial autocorrelation coefficient represents the spatial association between the two photovoltaic modules.

[0032] S104: Using the target meteorological influencing factors as global features, a graph neural network is used to adjust the temporal and spatial relationships in the graph structure. Based on the adjusted temporal and spatial relationships, the multi-source datasets are fused to obtain a fused feature set.

[0033] In this embodiment, the input data of the graph neural network is graph structure data, including a set of node features, a set of edges (temporal and spatial relationships), and target meteorological influencing factors; the output data is the adjusted set of edges. The graph neural network includes at least a fully connected layer, a convolutional layer, and a gated recurrent layer, which are connected sequentially. Other processing layers can be added between the three layers. This embodiment only describes the processing method containing only three layers; please refer to the embodiments below.

[0034] The training process of a graph neural network is as follows: The graph structure data, including a set of node features, a set of edges representing temporal and spatial relationships, and target meteorological influencing factors, is input into a graph neural network. The network sequentially performs initial mapping and feature transformation on the target meteorological influencing factors through fully connected layers, then aggregates neighbor node information and extracts spatial features based on the graph topology through convolutional layers, and finally models and dynamically updates temporal correlation features through gated recurrent layers. After multi-layer feature propagation, the network outputs an adjusted set of edges. In this embodiment, the loss function is constructed based on the difference between the output edge set and the true labels. The backpropagation algorithm is used to calculate the gradient of the learnable parameters of each layer in the network. The gradient optimization algorithm iteratively updates the parameters of the fully connected layers, convolutional layers, and gated recurrent layers, continuously reducing the error between the predicted and true results until the model converges, thus completing the training of the graph neural network.

[0035] In one embodiment, using the target meteorological influencing factor as a global feature, a graph neural network is used to adjust the temporal and spatial relationships in the graph structure, including: The time series of the target meteorological influencing factors are mapped to global embedding vectors through the fully connected layer of a graph neural network. The global embedding vectors are then concatenated into the power generation feature vector of each photovoltaic module node to obtain the enhanced node features. The convolutional layers of a graph neural network are used to perform spatial-dimensional message passing on the enhanced node features along the edges of the graph structure to update spatial relationships. By using the gated recurrent layer of a graph neural network, the node features under the updated spatial relationships are recursively analyzed in the time dimension to update the temporal relationships.

[0036] In this embodiment, the time series data of the target meteorological influencing factor is input into the first fully connected layer of the graph neural network and mapped to a global embedding vector. The dimension of this global embedding vector is the same as the dimension of the power generation feature vector of each node. The global embedding vector is concatenated to the power generation feature vector of each photovoltaic module node to obtain the enhanced node features. These enhanced node features not only contain their own historical power information but also incorporate meteorological background information, enabling the temporal and spatial correlations between subsequent nodes to perceive external meteorological conditions.

[0037] Furthermore, this embodiment utilizes the convolutional layers of a graph neural network to perform spatial message passing along the edges of the graph structure on the enhanced node features. The essence of message passing is to aggregate the enhanced features of each node's neighboring nodes and update its own feature representation. During the aggregation process, the edge weights (i.e., the strength of spatial relationships) may be learned or remain unchanged, but the changes in node features, in turn, reflect the degree of influence of spatial relationships on the final prediction. In this embodiment, based on the parameters of the graph convolutional layers of the trained graph convolutional neural network, the network can adaptively adjust the importance of information received from neighboring nodes, thereby updating spatial relationships.

[0038] Finally, this embodiment utilizes the gated recurrent layer of the graph neural network to perform temporal state recursion on the node features under the updated spatial correlation, enabling the network to selectively forget or retain power generation data from historical moments and integrate the current spatial aggregation features into the new hidden state, thereby updating the importance of power generation data from different historical moments to the prediction of the current moment, i.e., the temporal correlation.

[0039] This embodiment injects target meteorological influencing factors as global features into a graph neural network, enabling each photovoltaic module node to perceive a unified meteorological background during spatial message transmission, thus enhancing the global consistency of feature expression. Secondly, this embodiment utilizes graph convolutional layers to dynamically update spatial relationships along the graph structure, which can adaptively capture local spatial dependencies and microclimate differences between modules. Combined with gated recurrent layers for temporal state recursion, it effectively models the law of power evolution over time, ultimately achieving deep fusion of spatiotemporal features and improving the accuracy and robustness of power generation prediction.

[0040] In one embodiment, a fused feature set is obtained by fusing multi-source datasets based on adjusted temporal and spatial correlations, including: The adjusted temporal correlation is represented as the hidden state sequence of each node at different historical moments, and the adjusted spatial correlation is represented as the aggregated feature vector of each node and its neighboring nodes. Based on the cross-modal attention mechanism, the hidden state sequence is used as the query vector and the aggregated feature vector is used as the key-value pair to calculate the cross-attention weight between the time dimension and the spatial dimension. The hidden state sequence and aggregated feature vector are weighted and summed according to the cross-attention weights to obtain the weighted sum result. The weighted sum result is then residually connected with the global features of the target meteorological influencing factor to obtain the fused feature set.

[0041] In this embodiment, the adjusted temporal relationships are transformed into a sequence of hidden states of a single node at multiple historical moments through temporal modeling. This encodes the node's own evolution patterns, temporal dependencies, and dynamic trends over time, thus completing a structured representation of the temporal dimension information. The adjusted spatial relationships are then transformed into aggregated feature vectors of the node and its neighboring nodes through spatial graph aggregation. This encodes the node's neighborhood information, spatial correlation, and regional collaborative characteristics in the spatial topology, thus completing a structured representation of the spatial dimension information.

[0042] This embodiment employs a cross-modal attention mechanism to construct a spatiotemporal information interaction bridge: using the temporal hidden state sequence as the query vector and the spatial aggregated feature vector as the key-value pair, attention calculation automatically learns the cross-attention weights of the temporal and spatial dimensions. This allows for the filtering of spatial neighborhood information that significantly impacts the temporal changes of the current node and the suppression of redundant noise in the spatiotemporal dimensions. Based on the learned cross-attention weights, the temporal hidden state sequence and the spatial aggregated feature vector are weighted and then summed to obtain a preliminary spatiotemporal coupled fusion feature. This preliminary fusion feature retains temporal dynamism while incorporating spatial topological constraints, addressing the issues of feature redundancy and insufficient correlation mining caused by simple spatiotemporal information splicing. Finally, this embodiment performs a residual connection between the above preliminary fusion feature and the global features of the target meteorological influencing factors, outputting a multi-dimensional complementary feature set that simultaneously includes temporal dynamic dependence, spatial topological correlation, and global meteorological constraints.

[0043] This embodiment dynamically aligns features in the temporal and spatial dimensions through a cross-modal attention mechanism, enabling the graph neural network to adaptively learn which spatially located neighboring nodes are more important for predicting the current node at different historical moments, avoiding bias caused by manually setting fusion weights. Furthermore, this embodiment effectively alleviates the information decay problem in the network through residual connections, while directly incorporating target meteorological influencing factors into the fusion process, significantly improving the representational ability of the fused feature set, thereby enhancing the accuracy and generalization performance of photovoltaic power generation prediction.

[0044] S105: Based on the fusion feature set and photovoltaic power prediction model, the power generation prediction results for the second time period in the future are obtained.

[0045] In one embodiment, the photovoltaic power prediction model includes a first photovoltaic power prediction model and a second photovoltaic power prediction model; Based on the fused feature set and photovoltaic power prediction model, the power generation prediction results for the second time period are obtained, including: The first photovoltaic power prediction model is input into the fused feature set to obtain the first prediction result, and the second photovoltaic power prediction model is input into the fused feature set to obtain the second prediction result; the second photovoltaic power prediction model has a different structure from the first photovoltaic power prediction model; The first and second prediction results are weighted and fused to obtain the power generation prediction result for the second time period in the future.

[0046] In this embodiment, the first photovoltaic power prediction model can be a Long Short-Term Memory (LSTM) network model, and the second photovoltaic power prediction model can be a Transformer model.

[0047] Since a single model is prone to overfitting on the training set, and two models with different structures have different decision boundaries in the feature space, fusing them can cancel out their respective overfitting parts.

[0048] For example, the decision boundary characteristics of LSTM are as follows: influenced by recurrent structures, LSTM tends to learn sequential dependencies and long-term memory in time series, is sensitive to recursive relationships between time steps, but insensitive to correlations at distant locations. Its boundary is usually smoother and has some robustness to local noise. The decision boundary characteristics of Transformer models are as follows: based on a self-attention mechanism, Transformer models can directly capture the correlation between any two time steps, have no recurrent loops, and therefore can learn more complex, non-local dependency patterns.

[0049] In this embodiment, the first prediction result and the second prediction result are weighted and fused to obtain the power generation prediction result for the second time period in the future. This can be understood as the first prediction result being... The corresponding weighting coefficient is The second prediction result is The corresponding weighting coefficient is The predicted power generation result is: .

[0050] This embodiment uses different models to process the fused feature set, and reduces the bias of the single model by fusing the prediction results, thereby improving the accuracy of the prediction results.

[0051] As can be seen from the above, this embodiment filters target meteorological influencing factors by spatiotemporal correlation, eliminates redundant or irrelevant variables, and reduces input noise. Simultaneously, this embodiment constructs a graph structure using the geographical location of photovoltaic modules as nodes and spatiotemporal correlations as edges, and uses a graph neural network to dynamically adjust the relationships within the graph structure, enabling deep integration of power generation data, spatial feature data, and regional meteorological data. Compared to traditional single-model or simple feature overlay methods, this embodiment effectively captures the spatial dependence between photovoltaic modules and the influence of meteorology, significantly improving the accuracy and stability of power generation prediction.

[0052] In addition, this application embodiment extracts location and state features from spatial feature data and combines them with graph neural networks to adaptively adjust the spatiotemporal correlation, which can adapt to the layout differences of different photovoltaic power plants and the dynamic changes in meteorological conditions, thereby enhancing the model's adaptability to the complex and ever-changing actual operating environment.

[0053] In one embodiment of this application, the target meteorological impact factor includes at least one meteorological impact factor; Meteorological influencing factors were screened based on the spatiotemporal correlation between the power generation data of each photovoltaic module and regional meteorological data to obtain target meteorological influencing factors, including: Calculate the temporal and spatial correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data; If the temporal correlation is greater than the first threshold and the spatial correlation is greater than the second threshold, then the meteorological influence factor will be used as the target meteorological influence factor.

[0054] In this embodiment, for each photovoltaic module, a temporal correlation analysis is performed between its historical power generation sequence and the historical time series of a certain meteorological influencing factor to obtain the temporal correlation degree. This temporal correlation degree reflects the degree of direct impact of the meteorological influencing factor on the power generation of the photovoltaic module over time; for example, irradiance and power generation are usually positively correlated.

[0055] Considering the geographical differences of different photovoltaic modules, the spatial correlation between the observed values ​​of the meteorological influencing factor at different spatial locations and the power generation of the corresponding photovoltaic modules can be calculated. This embodiment can employ spatial autocorrelation methods or directly calculate the spatial covariance between the meteorological influencing factor and power generation at different photovoltaic modules. The spatial correlation reflects whether the spatial distribution of the meteorological influencing factor is consistent with the spatial pattern of power generation. For example, cloud cover can lead to a decrease in the power generation of upstream photovoltaic modules, and consequently, a decrease in the power generation of downstream photovoltaic modules, demonstrating a positive spatial correlation.

[0056] In this embodiment, the first threshold and the second threshold can be set based on experience. Only meteorological influencing factors that simultaneously satisfy the condition that the temporal correlation is greater than the first threshold and the spatial correlation is greater than the second threshold can be selected as target meteorological influencing factors.

[0057] This embodiment avoids introducing meteorological factors that are unrelated to or spuriously correlated with power changes by screening target meteorological influencing factors, thus reducing input noise to the model. The screened target meteorological influencing factors can also provide high-quality global features for subsequent edge adjustments in the graph neural network, improving the generalization ability and robustness of the prediction model.

[0058] In one embodiment, calculating the temporal correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data includes: The first number of partitions is determined based on the time series of power generation data for each photovoltaic module, and the second number of partitions is determined based on the time series of each meteorological influencing factor. Both the first number of partitions and the second number of partitions are positive integers. The grid division method is determined based on the first and second division numbers; Calculate the mutual information between the time series of power generation data for each photovoltaic module and the time series of each meteorological influencing factor under each grid partitioning method; Normalize each mutual information to obtain a normalized mutual information set, and use the maximum value in the normalized mutual information set as the temporal correlation between the power generation data of each photovoltaic module and each meteorological influencing factor.

[0059] In this embodiment, the first division number x represents the number of intervals into which the power generation data of the photovoltaic module is divided, and the second division number y represents the number of intervals into which the value range of any meteorological influencing factor is divided. Both the first division number x and the second division number y are positive integers.

[0060] In one embodiment, before determining the mesh partitioning method based on the first partition number and the second partition number, the method further includes: Determine the sample size corresponding to the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data within the first time period, and determine the upper limit function of the number of grids based on the sample size; The number of first partitions and the number of second partitions that satisfy the condition that the product of the number of first partitions and the number of second partitions is less than the upper limit function are taken as candidate partitions. The grid division method is determined based on the first number of divisions and the second number of divisions, including: The grid partitioning method is determined based on the number of candidate partitions.

[0061] In this embodiment, the sample size is set to n (i.e., the length of the time series), and an upper bound function is determined based on the sample size n. This value determines the upper limit of the product of the maximum allowed number of grid divisions x and y. For example, if data is collected every 15 minutes, the sample size for 30 days is n=2880. That is, all that satisfy All grid division methods will be taken into account.

[0062] In this embodiment, for all satisfying Given positive integers x and y, iterate through them one by one. Usually, x is fixed first, and y ranges from 2 to... (Round down) Iterate through the data; then increment x in turn. The values ​​of x and y determine how many grids the data points are divided into, thus allowing us to observe the relationships between variables at different granularities.

[0063] For each group The data space needs to be divided into The specific boundaries of each grid. There are two common methods: The first method is equal-frequency partitioning: the power generation data of photovoltaic modules are arranged in ascending order and divided into x equal segments, each containing as many data points as possible; the meteorological influence factors are also partitioned using equal-frequency partitioning. This partitioning method ensures that the number of data points in each cell is relatively balanced, avoiding too many empty cells due to uneven data distribution.

[0064] The second method is equidistant division: the grid is divided equidistantly according to the power generation data of photovoltaic modules and the value range of meteorological influence factors.

[0065] After the grid is generated, it is necessary to calculate the mutual information between the time series of power generation data for each photovoltaic module and the time series of each meteorological influencing factor under this grid generation method. Specifically: For the present Count the number of cells that fall into each grid. Number of data points The joint probability is calculated as follows: The marginal probability is , .

[0066] The formula for calculating mutual information is: .

[0067] Where X represents the power generation data of the photovoltaic module, and Y represents the value corresponding to the meteorological influence factor.

[0068] Divide the mutual information calculated above by the normalization factor. This yields the normalized mutual information value under this grid partitioning method: .

[0069] Iterate through all satisfied After combining (x, y), take all The maximum value in the coefficient is used as the final maximum information coefficient. The time correlation between the meteorological influencing factor and the power generation data of the photovoltaic module is the maximum information coefficient. The time correlation of other meteorological influencing factors can be calculated in the same way.

[0070] In this embodiment, since there may be a nonlinear saturation relationship between power generation data and irradiance, and a time lag relationship with temperature, the adaptive grid partitioning mechanism with the maximum information coefficient can effectively quantify these complex correlations, providing an accurate basis for subsequent spatiotemporal correlation screening.

[0071] In one embodiment of this application, the power generation data includes power generation values; Calculate the spatial correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data, including: For each meteorological influencing factor: Using the power generation value of photovoltaic modules as the dependent variable and the spatial interpolation of meteorological influencing factors at the geographical location of photovoltaic modules as the independent variable, a geographically weighted regression model is established, which includes local regression coefficients. A spatial weight matrix is ​​constructed based on the distance between every two photovoltaic modules; Local regression coefficients are calculated based on the spatial weight matrix and the geographic weighted regression model. Based on the local regression coefficients, the spatial correlation between meteorological influencing factors and the power generation data of each photovoltaic module is obtained.

[0072] In this embodiment, spatial interpolation means mapping meteorological influencing factors from a limited number of observation points to the location of each photovoltaic module, providing an estimated value of the meteorological influencing factor for each photovoltaic module, and thus using it as an independent variable in the geographically weighted regression model.

[0073] For each photovoltaic module a (a=1,...,m, where m is the total number of photovoltaic modules), the geographically weighted regression model takes the form of: ; in, It represents the local regression coefficient of meteorological influencing factors on power generation at the location of the photovoltaic module. In other words, it represents the average change in power generation when the meteorological influencing factors change by one unit. Its absolute value reflects the degree of spatial correlation. The larger the value, the stronger the degree of spatial correlation. The coordinates represent the geographical location of photovoltaic module a, with the horizontal and vertical axes respectively. This represents the spatial interpolation of meteorological influence factors at the geographical location of photovoltaic module a. This represents the power generation capacity of photovoltaic module a. Indicates random error. It represents the local intercept, reflecting the base power level.

[0074] In this embodiment, the geographically weighted regression model can quantify the local influence intensity of each meteorological influencing factor at different photovoltaic modules, thereby extracting spatial correlation. Compared with traditional global regression models, it can better reflect the spatial heterogeneity within photovoltaic sites caused by factors such as terrain, shadows, and microclimates.

[0075] In this embodiment, the spatial weight matrix is ​​an m×m matrix, where each element represents the weight of the observation data from other photovoltaic modules at the current photovoltaic module. This weight is typically calculated based on the distance between photovoltaic modules; closer modules have a higher weight, and farther modules have a lower weight. This distance can be Euclidean distance or Manhattan distance.

[0076] The local regression coefficients are calculated based on the spatial weight matrix and the geographic weighted regression model, including: For each photovoltaic module a, its local regression coefficient (T represents transpose) can be solved using weighted least squares: ; Where X is the matrix of independent variables for all samples, and Y is the vector of dependent variables for all samples. This is a spatial weighted diagonal matrix constructed with photovoltaic module a as the center.

[0077] The essence of the above formula is to find the formula that minimizes the weighted error while minimizing the sum of squared weighted residuals. .

[0078] In one embodiment of this application, the spatial correlation between meteorological influence factors and the power generation data of each photovoltaic module is obtained based on local regression coefficients, including: If the local regression coefficient passes the significance test, the absolute value of the local regression coefficient that passes the significance test will be used as the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module. If the local regression coefficient fails the significance test, the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module will be set to zero.

[0079] In this embodiment, to ensure the statistical reliability of the spatial correlation, a significance test, such as a t-test, is typically performed on each local regression coefficient. Only local regression coefficients with a significance test probability less than 0.05 are considered statistically significant, and only then can the absolute value of the local regression coefficient that passes the significance test be used as the spatial correlation between the meteorological influence factor and the power generation data of the corresponding photovoltaic module.

[0080] If the local regression coefficient fails the significance test, it is considered that the local regression coefficient is not significantly different from 0, that is, the meteorological influencing factor has no significant spatial influence on the photovoltaic module.

[0081] Corresponding to the renewable energy power generation prediction method in the above embodiments, Figure 2 This is a structural block diagram of a renewable energy power generation prediction device according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The renewable energy power generation prediction device 20 is applied to the processor in the photovoltaic power generation system, which also includes multiple photovoltaic modules. The renewable energy power generation prediction device 20 includes: a data acquisition module 21, a data filtering module 22, a graph structure construction module 23, a data fusion module 24, and a prediction module 25.

[0082] Among them, the data acquisition module 21 is used to acquire multi-source datasets, which include power generation data and spatial feature data of each photovoltaic module within the first historical time period, as well as regional meteorological data of the area where the photovoltaic power generation system is located; the spatial feature data represents the location and status characteristics of the photovoltaic modules in the spatial dimension; The data filtering module 22 is used to filter meteorological influencing factors based on the spatiotemporal correlation between the power generation data of each photovoltaic module and regional meteorological data, so as to obtain the target meteorological influencing factors. Graph structure construction module 23 is used to construct a graph structure with the geographical location of each photovoltaic module as nodes and the temporal and spatial correlations between the spatial feature data and the power generation data of each photovoltaic module as edges. The data fusion module 24 is used to take the target meteorological influencing factor as the global feature, use a graph neural network to adjust the temporal and spatial relationships in the graph structure, and fuse the multi-source datasets based on the adjusted temporal and spatial relationships to obtain a fused feature set. Prediction module 25 is used to obtain the power generation prediction results for the second time period in the future based on the fused feature set and the photovoltaic power prediction model.

[0083] In one embodiment of this application, the data fusion module 24 is specifically used for: The adjusted temporal correlation is represented as the hidden state sequence of each node at different historical moments, and the adjusted spatial correlation is represented as the aggregated feature vector of each node and its neighboring nodes. Based on the cross-modal attention mechanism, the hidden state sequence is used as the query vector and the aggregated feature vector is used as the key-value pair to calculate the cross-attention weight between the time dimension and the spatial dimension. The hidden state sequence and aggregated feature vector are weighted and summed according to the cross-attention weights to obtain the weighted sum result. The weighted sum result is then residually connected with the global features of the target meteorological influencing factor to obtain the fused feature set.

[0084] In one embodiment of this application, the target meteorological influence factor includes at least one meteorological influence factor. The data filtering module 22 is specifically used for: Calculate the temporal and spatial correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data; If the temporal correlation is greater than the first threshold and the spatial correlation is greater than the second threshold, then the meteorological influence factor will be used as the target meteorological influence factor.

[0085] In one embodiment of this application, the data filtering module 22 is specifically used for: The first number of partitions is determined based on the time series of power generation data for each photovoltaic module, and the second number of partitions is determined based on the time series of each meteorological influencing factor. Both the first number of partitions and the second number of partitions are positive integers. The grid division method is determined based on the first and second division numbers; Calculate the mutual information between the time series of power generation data for each photovoltaic module and the time series of each meteorological influencing factor under each grid partitioning method; Normalize each mutual information to obtain a normalized mutual information set, and use the maximum value in the normalized mutual information set as the temporal correlation between the power generation data of each photovoltaic module and each meteorological influencing factor.

[0086] In one embodiment of this application, the renewable energy power generation prediction device 20 further includes a grid division module, used to determine the sample size corresponding to the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data within a first time period, and to determine an upper limit function for the number of grids based on the sample size; The number of first partitions and the number of second partitions that satisfy the condition that the product of the first partition number and the second partition number is less than the upper limit function are taken as candidate partition numbers.

[0087] The data filtering module 22 is specifically used to determine the grid division method based on the number of candidate divisions.

[0088] In one embodiment of this application, the power generation data includes power generation values; the data filtering module 22 is specifically used for: For each meteorological influencing factor: Using the power generation value of photovoltaic modules as the dependent variable and the spatial interpolation of meteorological influencing factors at the geographical location of photovoltaic modules as the independent variable, a geographically weighted regression model is established, which includes local regression coefficients. A spatial weight matrix is ​​constructed based on the distance between every two photovoltaic modules; Local regression coefficients are calculated based on the spatial weight matrix and the geographic weighted regression model. Based on the local regression coefficients, the spatial correlation between meteorological influencing factors and the power generation data of each photovoltaic module is obtained.

[0089] In one embodiment of this application, the data filtering module 22 is specifically used for: If the local regression coefficient passes the significance test, the absolute value of the local regression coefficient that passes the significance test will be used as the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module. If the local regression coefficient fails the significance test, the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module will be set to zero.

[0090] See Figure 3 , Figure 3 This is a schematic block diagram of a photovoltaic power generation system provided in an embodiment of this application. Figure 3 The photovoltaic power generation system 300 shown in this embodiment may include: one or more processors 301, one or more input devices 302, one or more output devices 303, one or more memories 304, and multiple photovoltaic modules 305. The processors 301, input devices 302, output devices 303, memories 304, and photovoltaic modules 305 communicate with each other via a communication bus 306. The memory 304 stores computer programs, including program instructions. The processor 301 executes the program instructions stored in the memory 304. Specifically, the processor 301 is configured to invoke the program instructions to perform the functions of each module in the above-described device embodiments, for example... Figure 2 The functions of the data acquisition module 21, data filtering module 22, graph structure construction module 23, data fusion module 24, and prediction module 25 are shown.

[0091] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), but it may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0092] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0093] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0094] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the renewable energy power generation prediction method provided in the embodiments of this application, or they can execute the implementation method of the photovoltaic power generation system described in the embodiments of this application, which will not be repeated here.

[0095] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0096] The computer-readable storage medium can be an internal storage unit of the photovoltaic power generation system in any of the foregoing embodiments, such as a hard drive or memory of the photovoltaic power generation system. The computer-readable storage medium can also be an external storage device of the photovoltaic power generation system, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the photovoltaic power generation system. Furthermore, the computer-readable storage medium can include both internal storage units and external storage devices of the photovoltaic power generation system. The computer-readable storage medium is used to store computer programs and other programs and data required by the photovoltaic power generation system. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0097] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0098] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the photovoltaic power generation system and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed photovoltaic power generation system and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or modules, or it may be an electrical, mechanical, or other form of connection.

[0100] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0101] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0102] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting renewable energy power generation, characterized in that, A processor used in a photovoltaic power generation system, the photovoltaic power generation system further including multiple photovoltaic modules; the method includes: A multi-source dataset is obtained, which includes power generation data and spatial feature data of each photovoltaic module within a historical first time period, as well as regional meteorological data of the area to which the photovoltaic power generation system belongs; the spatial feature data represents the location and state characteristics of the photovoltaic modules in the spatial dimension. Meteorological influencing factors are screened based on the spatiotemporal correlation between the power generation data of each photovoltaic module and the regional meteorological data to obtain target meteorological influencing factors; A graph structure is constructed using the geographical location of each photovoltaic module as nodes and the temporal and spatial correlations between the spatial feature data and the power generation data of each photovoltaic module as edges. Using the target meteorological influencing factor as a global feature, a graph neural network is used to adjust the temporal and spatial relationships in the graph structure. Based on the adjusted temporal and spatial relationships, the multi-source dataset is fused to obtain a fused feature set. Based on the fusion feature set and the photovoltaic power prediction model, the power generation prediction results for the second time period in the future are obtained.

2. The renewable energy power generation prediction method as described in claim 1, characterized in that, The process of fusing the multi-source datasets based on the adjusted temporal and spatial correlations to obtain a fused feature set includes: The adjusted temporal correlation is represented as the hidden state sequence of each node at different historical moments, and the adjusted spatial correlation is represented as the aggregated feature vector of each node and its neighboring nodes. Based on the cross-modal attention mechanism, the hidden state sequence is used as the query vector, and the aggregated feature vector is used as the key-value pair to calculate the cross-attention weight between the time dimension and the spatial dimension. The hidden state sequence and the aggregated feature vector are weighted and summed according to the cross-attention weights to obtain a weighted sum result. The weighted sum result is then residually concatenated with the global features of the target meteorological influencing factor to obtain the fused feature set.

3. The renewable energy power generation prediction method as described in claim 1, characterized in that, The target meteorological influencing factors include at least one meteorological influencing factor; The meteorological influencing factors are screened based on the spatiotemporal correlation between the power generation data of each photovoltaic module and the regional meteorological data to obtain target meteorological influencing factors, including: Calculate the temporal and spatial correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data; If the temporal correlation is greater than the first threshold and the spatial correlation is greater than the second threshold, then the meteorological influence factor is taken as the target meteorological influence factor.

4. The renewable energy power generation prediction method as described in claim 3, characterized in that, The calculation of the time correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data includes: The first number of partitions is determined based on the time series of power generation data for each photovoltaic module, and the second number of partitions is determined based on the time series of each meteorological influencing factor. Both the first number of partitions and the second number of partitions are positive integers. The grid division method is determined based on the first number of divisions and the second number of divisions; Calculate the mutual information between the time series of power generation data for each photovoltaic module and the time series of each meteorological influencing factor under each grid partitioning method; Each mutual information is normalized to obtain a normalized mutual information set, and the maximum value in the normalized mutual information set is used as the time correlation between the power generation data of each photovoltaic module and each meteorological influencing factor.

5. The renewable energy power generation prediction method as described in claim 4, characterized in that, Before determining the grid division method based on the first number of divisions and the second number of divisions, the method further includes: Determine the sample size corresponding to the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data within the first time period, and determine the upper limit function of the number of grids based on the sample size; The first and second partition numbers that satisfy the condition that the product of the first partition number and the second partition number is less than the upper limit function are taken as candidate partition numbers. The process of determining the grid division method based on the first number of divisions and the second number of divisions includes: The grid partitioning method is determined based on the number of candidate partitions.

6. The renewable energy power generation prediction method as described in claim 3, characterized in that, The power generation data includes power generation values; The calculation of the spatial correlation between the power generation data of each photovoltaic module and each meteorological influencing factor in the regional meteorological data includes: For each meteorological influencing factor: Using the power generation value of the photovoltaic module as the dependent variable and the spatial interpolation of the meteorological influencing factor at the geographical location of the photovoltaic module as the independent variable, a geographically weighted regression model is established, which includes local regression coefficients. A spatial weight matrix is ​​constructed based on the distance between every two photovoltaic modules; The local regression coefficients are calculated based on the spatial weight matrix and the geographically weighted regression model, and the spatial correlation between the meteorological influencing factors and the power generation data of each photovoltaic module is obtained based on the local regression coefficients.

7. The renewable energy power generation prediction method as described in claim 6, characterized in that, The spatial correlation between the meteorological influencing factors and the power generation data of each photovoltaic module, obtained based on the local regression coefficients, includes: If the local regression coefficient passes the significance test, the absolute value of the local regression coefficient that passes the significance test will be used as the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module. If the local regression coefficient fails the significance test, the spatial correlation between the meteorological influencing factor and the power generation data of the corresponding photovoltaic module is set to zero.

8. A renewable energy power generation prediction device, characterized in that, A processor for use in a photovoltaic power generation system, the photovoltaic power generation system also including multiple photovoltaic modules; the device includes: The data acquisition module is used to acquire a multi-source dataset, which includes power generation data and spatial feature data of each photovoltaic module within a historical first time period, as well as regional meteorological data of the area to which the photovoltaic power generation system belongs; the spatial feature data represents the location and state characteristics of the photovoltaic modules in the spatial dimension. The data filtering module is used to filter meteorological influencing factors based on the spatiotemporal correlation between the power generation data of each photovoltaic module and the regional meteorological data, so as to obtain the target meteorological influencing factors. The graph structure construction module is used to construct a graph structure with the geographical location of each photovoltaic module as nodes and the temporal and spatial correlations between the spatial feature data and the power generation data of each photovoltaic module as edges. The data fusion module is used to take the target meteorological influencing factor as a global feature, use a graph neural network to adjust the temporal and spatial relationships in the graph structure, and fuse the multi-source dataset based on the adjusted temporal and spatial relationships to obtain a fused feature set. The prediction module is used to obtain the power generation prediction result for the second time period in the future based on the fused feature set and the photovoltaic power prediction model.

9. A photovoltaic power generation system, comprising a plurality of photovoltaic modules, a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images