Regional distributed photovoltaic power prediction method and system based on multi-element data cross-modal fusion

Abstract

The application relates to a regional distributed photovoltaic power prediction method and system based on multi-element data cross-modal fusion, and belongs to the technical field of photovoltaic power prediction. First, through a dynamic perception convolution kernel of an AG-CNN, spatial characteristics of meteorological data and spatial heterogeneity of power data are adaptively extracted, so as to provide a high-quality spatial basis for cross-modal fusion. Secondly, a double-dimension dynamic attention mechanism of a DR-Transformer is used to deeply mine cross-modal correlations of 'historical power time series-future meteorological driving', dynamically allocate weights to adapt to fusion requirements in different scenes, and capture correlation evolution in a long period. Thirdly, a cross-modal feature bridging module is designed. Finally, the robustness of the model is verified in typical scenes to ensure that the model meets the accuracy requirements of power grid multi-scale scheduling.

Description

Technical Field

[0001] This invention relates to a method and system for predicting regional distributed photovoltaic power based on multi-data cross-modal fusion, belonging to the field of photovoltaic power prediction technology. Background Technology

[0002] Current research on distributed photovoltaic power prediction has developed multiple technical approaches, but existing methods still face limitations in both "power-meteorological cross-modal fusion" and "long-term time-series dependency modeling," which consistently restricts the improvement of prediction accuracy. From the perspective of physical models, these models are built based on the conversion mechanism of photovoltaic modules. Although they can reflect the direct impact of meteorological factors on power and show a certain degree of reliability in short-term predictions, they rely on fixed module parameters and high-resolution meteorological data. They are not only unable to characterize the long-term cumulative effects of meteorological factors (such as cloud cover and temperature), but also lack dynamic cross-modal adaptation of power and meteorological data. Simply linking the two through mechanism formulas cannot meet the fusion requirements under scenarios of meteorological abrupt changes or heterogeneous data.

[0003] The limitations of statistical models lie in their shallow cross-modal correlations and insufficient adaptation to nonlinearity. These models establish correlations by mining patterns in historical data and can achieve basic predictions when the data distribution is stable. However, they not only fail to capture the nonlinear relationship between meteorological data and power, but their utilization of cross-modal information is also limited to shallow statistical correlations (such as simple linear regression). When the modal differences between meteorological and power data increase (such as the sudden drop in irradiance during cloudy and rainy weather leading to increased power fluctuations), the model's characterization of the "meteorological drive-power response" correlation quickly becomes distorted, and its accuracy drops sharply.

[0004] These limitations are not isolated, but rather intertwined and progressively reinforce each other around the two core issues of cross-modal fusion and long-term time-series dependence, which together constitute the modeling difficulties of distributed photovoltaic power prediction. The former determines whether meteorological and power data can be effectively coordinated, while the latter affects whether this coordination relationship can be stably transmitted over time. The coupling of the two further amplifies the modeling complexity.

[0005] The aforementioned challenges, when combined, ultimately point to a bottleneck in the top-level architecture: insufficient time-space-cross-modal collaborative modeling capabilities. On the one hand, most models fragment spatial correlations and cross-modal fusion; for example, some spatiotemporal models only capture station spatial correlations at the power level, failing to incorporate meteorological spatial distribution into the collaborative consideration. On the other hand, traditional time-series architectures struggle to support long-term cross-modal correlation transmission: LSTM memory units decay over time, easily losing early key meteorological features; and Transformer attention is easily distracted in long sequences, making it difficult to focus on the core period of "meteorological trend - power trend." This lack of collaboration makes it difficult for existing models to balance regional differences and long-term coupling, severely restricting the generalization ability and robustness of prediction models under complex meteorological conditions. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a regional distributed photovoltaic (PV) power prediction method and system based on multi-data cross-modal fusion. With accurate distributed PV power prediction as its core objective, it focuses on the synergistic breakthrough of deep cross-modal fusion of power and meteorological data and accurate capture of long-term time-series dependencies. Combining the technical characteristics of Adaptive Spatial Convolution (AG-CNN) and Dynamic Relation Transformer (DR-Transformer), an end-to-end combined prediction framework is constructed. First, the dynamic perceptual convolution kernel of AG-CNN achieves the adaptation and extraction of spatial features of meteorological data and spatial heterogeneity of power data—preserving the continuous structure of meteorological data while aligning with the "numerous points and wide distribution" characteristics of distributed PV, providing a high-quality spatial foundation for cross-modal fusion. Second, utilizing the dual-dimensional dynamic attention mechanism of DR-Transformer, it deeply mines the cross-modal correlation of "historical power time series - future meteorological drivers," adapting to the fusion requirements of different scenarios through dynamic weight allocation, while capturing the correlation evolution over long periods. Third, a cross-modal feature bridging module is designed to specifically address the interaction between the convolutional features output by AG-CNN and DR-Transformer. The heterogeneity of the input sequence features in terms of dimension and semantics enables seamless collaboration between spatial features and cross-modal temporal features. Finally, the robustness of the model is verified in typical scenarios to ensure that it meets the accuracy requirements of multi-scale grid scheduling. This invention aims to overcome the dual limitations of existing models in cross-modal fusion and long-term time-series modeling, providing a more reliable technical solution for distributed photovoltaic power prediction, thereby supporting the safe and economical operation of high-penetration photovoltaic grid-connected power grids, and possessing both theoretical depth and practical engineering value.

[0007] The technical solution of the present invention is as follows:

[0008] The regional distributed photovoltaic power prediction method based on multivariate data cross-modal fusion has the following steps:

[0009] (1) Spatiotemporal data gridding consistency processing: collect the original meteorological data and historical power data, and standardize the data. Then, based on the unified geographic coordinates, realize the spatial correlation between meteorological data and historical power data.

[0010] (2) Based on the spatial features of AG-CNN with dynamic perception convolution, the data is adaptively extracted. AG-CNN relies on a unified basic spatial benchmark and achieves refined local modeling through dynamic feature perception and cross-regional information fusion. Its core is to build a feature extraction mechanism that has both spatial consistency and flexible adaptability. It can ensure the natural correlation between meteorological and power data in the spatial dimension and accurately capture the feature pattern differences in different regions.

[0011] (3) Cross-modal feature bridging module design: Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, the high-dimensional local features output by AG-CNN are transformed into a spatiotemporal token sequence that can be parsed by DR-Transformer, while retaining key spatial details and cross-regional correlation information;

[0012] (4) DR-Transformer time-series correlation modeling with dual-dimensional dynamic attention. DR-Transformer addresses the core limitations of existing models such as "insufficient capture of long-term time-series dependencies" and "loose coupling of cross-modal features". It takes the spatial dimension + temporal dimension dynamic attention mechanism as the core, deeply integrates the long-term time-series dependency of historical power with the driving role of future meteorological data, realizes the accurate coupling of meteorological-power cross-modal spatiotemporal features, and obtains the total power prediction result of distributed photovoltaic system. Its design goal is to solve the problems of "transmission of meteorological uncertainty" and "attenuation of long-period time-series correlation", so that the model can capture the long-term evolution law of historical power and accurately respond to the dynamic driving of future meteorological conditions.

[0013] According to a preferred embodiment of the present invention, in step (1), the meteorological data includes solar irradiance, temperature, and humidity, etc.

[0014] The data standardization process processes meteorological data and historical power data separately to eliminate the influence of different characteristic units. The Z-score standardization method is used, and the formula is as follows:

[0015] (1)

[0016] Where x represents the original data, μ represents the data mean, and σ represents the standard deviation. This is the standardized data.

[0017] Spatial correlation: Combining the coverage of photovoltaic arrays and meteorological monitoring points, and based on the spatial scale of meteorological data, the geographical location of photovoltaic panels is matched with the corresponding meteorological regions, so that power data and meteorological data form a gridded spatial correlation. At the same time, the data is time-aligned according to the time axis of prediction granularity, and abnormal data is removed, forming a spatiotemporally aligned input structure, which lays the foundation for subsequent models to capture spatiotemporal correlations.

[0018] According to a preferred embodiment of the present invention, in step (2), the core design of AG-CNN is dynamic perceptual convolutional kernel and adaptive region fusion: with grid cells as the basic spatial index, the convolutional kernel of AG-CNN adapts in real time to the station distribution density and feature fluctuation pattern in different grid cells through learnable parameters. For example, in grid cells with dense stations, the kernel function automatically focuses on local details (irradiance gradient, cloud shadow edge and other subtle features) in the cell; in grid cells with sparse stations, the kernel function flexibly extends to adjacent grid cells to capture a wider range of meteorological trends (such as cloud movement at the regional scale). This dynamic adjustment mechanism enables the model to naturally fit the actual spatial feature distribution and avoid mechanical dependence on fixed grid boundaries.

[0019] The convolutional kernel weights of AG-CNN are constructed collaboratively by a Gaussian function offset by spatial relative positions and learnable parameters.

[0020] (2)

[0021] Where (i, j) represents the spatial relative position offset, such as the offset of the surrounding area centered on the target area, i is the horizontal spatial relative position offset, j is the vertical spatial relative position offset, and σ g θ is the Gaussian kernel bandwidth, used to control the decay rate of the spatial influence range; s(i,j) is the feature similarity between the offset region and the target region, including information such as the consistency of power fluctuation patterns and irradiance change trends; θ is a learnable parameter, α=5, used to balance the influence of spatial location correlation and feature pattern similarity. When performing convolution operations on the feature map, the output features will adaptively fuse information from the target region and surrounding related regions.

[0022] (3)

[0023] Among them, y x，y To output the feature values ​​of the target region in the feature map, x x+i，y+j For features of surrounding related regions in the input feature map, such as irradiance, temperature, and average power within the region, the model can maintain natural spatial continuity and capture feature relationships between different regions through dynamic weights w(i,j), providing a feature foundation that combines spatial consistency and local detail for subsequent global modeling.

[0024] The AG-CNN network architecture employs a multi-layer convolutional block cascade design: the first layer captures instantaneous irradiance abrupt changes within a region using a dynamic kernel; the second layer extracts cross-regional meteorological propagation patterns using a dynamic kernel; and the third layer fuses features from the first two layers through convolution, enhancing key details. Each convolutional layer undergoes batch normalization and activation processing to enhance feature discriminative power and mitigate the gradient vanishing problem. This design allows the convolutional kernel to dynamically adjust its activation intensity based on regional features as it slides, effectively addressing the difficulty of adapting fixed kernels to complex spatial distributions.

[0025] According to a preferred embodiment of the present invention, in step (3), the cross-modal feature bridging module (CMFB) adapts the AG-CNN features to the DR-Transformer input. The feature map output by AG-CNN is first flattened into a one-dimensional sequence according to the region index, and then the channel dimension is mapped to the hidden dimension of DR-Transformer through linear projection to form an initial token sequence. To balance the preservation of local features and the adaptability of global modeling, CMFB dynamically adjusts the feature weights through a gating mechanism: for regions with high photovoltaic capacity and large power fluctuations, the feature weights are strengthened; for regions without sites or with stable features, noise interference is weakened. At the same time, the temporal features of historical power are introduced, and the spatial features and temporal context are explicitly fused through the spatiotemporal gating unit (STGU) to generate a token sequence that carries both regional details and temporal correlation, ensuring a smooth transition of features from the convolutional feature map to the Transformer sequence.

[0026] The feature map F output by AG-CNN is a four-dimensional tensor:

[0027] (4)

[0028] Where B is the training batch size, C is the number of feature channels, and H×W corresponds to the number of region divisions;

[0029] First, the two-dimensional region structure is transformed into a one-dimensional sequence through spatial flattening, and the channel dimension is adapted to the hidden dimension of the Transformer through learnable linear transformation:

[0030] (5)

[0031] in, The flattened sequence features are represented by N = H × W, where N is the total number of regions. Let be the projection matrix. Here, D is the bias term and D is the hidden dimension of the Transformer. This step compresses the local features of each region into tokens of fixed dimensions, forming a sequence of length N, which provides the basic unit for global modeling.

[0032] Since the optimization objectives of AG-CNN and DR-Transformer differ, a reparameterization layer is introduced to dynamically adjust the feature distribution to alleviate training bias:

[0033] (6)

[0034] in, For element-wise gating coefficients, It is the Sigmoid activation function. , As learnable parameters, the gating mechanism achieves dynamic alignment of feature distributions between the two models by strengthening key features and suppressing noise;

[0035] To display the temporal features of fused historical power and the spatial features of AG-CNN, a spatiotemporal gating unit (STGU) is introduced, and the correlation weights between the two are dynamically calibrated through cross-modal attention:

[0036] Let the historical power sequence be , among which, T h Let R be the historical time step, B be the training batch size, N be the total number of regions, and D be the Transformer hidden dimension. Then the gated fusion process is as follows:

[0037] (7)

[0038] in, F is the dynamic fusion coefficient. in The output features of the Space-Time Gated Unit (STGU) carry both spatial details and temporal correlations. The STGU is a space-time gated unit used to fuse spatial and temporal features. α For the cross-modal attention weight matrix, b α For cross-modal attention bias term, F reparam The reparameterized spatial features output by equation (6), As a cross-modal attention mechanism, it achieves semantic-level calibration of spatial and temporal features by learning the association weights between them.

[0039] Final output Directly used as input to the DR-Transformer encoder, each token retains both the local spatial details extracted by AG-CNN and incorporates the temporal context of historical power, providing a high-quality feature foundation for the subsequent dynamic attention mechanism to accurately capture the dual dependence of "spatial correlation-temporal evolution".

[0040] According to a preferred embodiment of the present invention, in step (4), the DR-Transformer serves as the core of global spatiotemporal fusion and realizes cross-modal correlation modeling of historical power and future weather through an encoder-decoder structure;

[0041] The encoder receives the sequence output from the cross-modal feature bridging module, processes historical power and spatial features, and captures implicit correlations and long-term temporal dependencies between regions through dynamic spatial attention. It employs a multi-layered stacked structure: the bottom layer introduces a dynamic window partitioning strategy, strengthening the temporal correlation of local regions through attention calculation within the window; the upper layer uses global attention to capture long-distance temporal dependencies. Each layer contains multi-head dynamic spatial attention and a feedforward neural network, where the attention weights are modulated by both the relative position of the regions and the similarity of power features. For example, for regions with high historical power correlation coefficients, the attention ratio between them is automatically increased.

[0042] The decoder processes meteorological data for future time steps. Under the constraint of the masked self-attention mechanism, it dynamically couples meteorological features with historical features output by the encoder through cross-attention. It adopts a multi-layer stacked structure, with each layer consisting of multi-head masked self-attention, multi-head cross-attention, and a feedforward neural network: masked self-attention ensures the temporal causality of the meteorological sequence; cross-attention adjusts the weights through dynamic correlation coefficients, for example, strengthening the attention ratio of irradiance during noon and increasing the driving weight of cloud cover during cloudy periods. After mapping the encoder and decoder features to the same dimension through a linear projection layer, cross-attention achieves a precise correlation between historical power trends and future meteorological drivers, outputting features that fuse global spatiotemporal information.

[0043] Finally, the output layer converts the fused features into regional power prediction values ​​for multiple future time steps through a fully connected layer. Then, based on the photovoltaic installed capacity of each region, the values ​​are weighted and summed to obtain the total power prediction result of the distributed photovoltaic system, completing the entire process from input data to prediction output.

[0044] According to a preferred embodiment of the present invention, the core design of DR-Transformer focuses on a dynamic attention mechanism in both spatial and temporal dimensions: in the spatial dimension, attention weights are adaptively adjusted based on the similarity of regional features and relative positional relationships to avoid mechanical dependence on fixed spatial units; in the temporal dimension, the evolution of historical power and the driving effect of future weather are dynamically coupled through a cross-attention mechanism to achieve deep fusion of cross-modal features. Its design focus is not on following a fixed structure, but on how to make the correlation modeling of spatial and temporal features more consistent with the actual fluctuation patterns of distributed photovoltaics through the dynamic adjustment of the attention mechanism.

[0045] To address the discrete distribution characteristics of distributed photovoltaic (PV) sites, the DR-Transformer encoder captures implicit correlations between regions through dynamic spatial attention. The attention weight calculation considers both relative spatial distance and feature pattern similarity. The formula for calculating the dynamic spatial attention weight is as follows:

[0046] (8)

[0047] Where q, k, and v are the query, key, and value vectors, respectively, and d k For feature dimension, β(r) is the feature similarity coefficient (calculated based on the cosine similarity of the regional power sequence, with a value of 0-1), and β(r) is the relative spatial distance coefficient, which is generated based on the Gaussian decay function and reflects the spatial correlation between regions. Through the coordinated modulation of the two, the attention weight can not only reflect the natural continuity in space, but also focus on the related regions with similar features, avoiding the correlation fragmentation caused by fixed spatial division.

[0048] The DR-Transformer's decoder dynamically fuses future meteorological data with historical features from the encoder output, achieving cross-modal correlation through masked self-attention and cross-attention.

[0049] Future meteorological data has a natural time-series dependency, but directly using self-attention may lead to future information being focused on. Masked self-attention forces the model to follow temporal causality through the lower triangular matrix, allowing only the current time step to focus on historical and concurrent meteorological characteristics.

[0050] Let the future weather sequence be the input to the decoder. T represents the number of future prediction steps, D represents the dimension of the meteorological feature vector, R represents the real number field, and m t Given the meteorological feature vector at step t, a query is generated through linear projection. ,key ,value , , , , where is the meteorological feature projection matrix, and h is the number of attention heads;

[0051] For time step t, the similarity of the mask self-attention is calculated as follows:

[0052] (9)

[0053] in, For the time step that is being noticed, This is a scaling factor to prevent the softmax gradient from vanishing due to excessively large inner product values.

[0054] Introducing a lower triangular mask matrix Suppressing future information through masking:

[0055] (10)

[0056] In the formula, This represents the t-th row and the th element in the mask matrix. The elements of the column, MaskedSim is the attention similarity score after masking;

[0057] Final The masked self-attention output MaskedAttn(t) of the step is:

[0058] (11)

[0059] in, The output projection matrix is ​​given by v, which represents the value vector. softmax(·) is the normalization exponential function, and Concat(·) represents the multi-head attention concatenation operation. After concatenation and projection, the meteorological features containing time-dependent characteristics are obtained from the multi-head attention results. .

[0060] According to a preferred embodiment of the present invention, in distributed photovoltaic data, the contribution of different regions to the total power varies significantly, and the power prediction needs to meet the requirements of overall fitting accuracy and spatiotemporal correlation. A single loss function is difficult to take into account these requirements. Therefore, a multi-objective loss function is designed to collaboratively optimize the model parameters, which not only retains the role of spatial benchmark in integrating discrete data, but also enhances the model's ability to capture dynamic meteorological-power correlation.

[0061] As the core foundation of the loss function, it is necessary to prioritize ensuring the model's fitting accuracy to the overall power level. Considering that photovoltaic power prediction is a continuous value regression problem, and that the penalty for larger errors is more stringent in practical applications, the squared error penalty characteristic of MSE can amplify the impact of extreme biases, forcing the model to focus on reducing large power errors. At the same time, its continuous differentiability is compatible with the model's gradient descent optimization mechanism, ensuring the stability of the training process. Therefore, the method constructs a multi-objective loss function oriented towards regional characteristics and cross-modal constraints:

[0062] The mean squared error (MSE) is used as the basic loss term, and the formula is as follows:

[0063] (12)

[0064] Where N is the total number of regions. To predict the time step, and Let be the predicted power and the actual power of the i-th region at time t, respectively;

[0065] AG-CNN extracts local features through dynamic kernels and cross-regional fusion, but its output regional features may suffer from fragmented correlations between adjacent regions due to the randomness of dynamic neighborhoods. The core of the regional feature consistency loss is to force the model to respect the implicit correlations between regions in physical space, ensuring that feature extraction not only focuses on the local details of a single region, but also captures the collaborative changes in adjacent regions caused by common meteorological disturbances.

[0066] (13)

[0067] Where, N i For the dynamic neighborhood of region i, w i，j The feature similarity weights for regions i and j are as follows: the higher the similarity, the greater the weight. This term compensates for the spatial correlation fragmentation problem caused by fixed grid division by constraining the consistency of power differences between adjacent regions.

[0068] The core of DR-Transformer is to fuse historical power time-series features with future meteorological driving features through cross-attention. However, the model may exhibit prediction bias due to an unreasonable allocation of attention weights, resulting in a "disconnect between meteorological and power trends." The goal of cross-modal time-series correlation loss is to force the model to learn the physical causal relationship between meteorological factors and power changes, ensuring that cross-modal correlations in the time-series dimension conform to objective laws, rather than merely fitting the surface correlation of data. It regulates the model's correlation learning from the physical mechanism level through dual constraints: trend consistency + reasonable attention weights.

[0069] (14)

[0070] in, As a trend consistency loss function, it is used to constrain the physical rationality of the direction of change. β is the weighting coefficient of the trend consistency loss, used to balance the strength of trend constraint and attention weight constraint.

[0071] (15)

[0072] in, To predict the time-series variation of power, This represents the temporal variation of meteorological factors. For cross-entropy loss, a high penalty is imposed for situations where the power trend is opposite to the meteorological trend;

[0073] As a cross-modal attention loss function, it is used to constrain the objective correlation of attention weights:

[0074] (16)

[0075] in, In the DR-Transformer cross-attention, meteorological features affect the region. exist Attention weight at any moment This is the true correlation coefficient between meteorology and power calculated based on historical data.

[0076] The overall loss function is the weighted sum of the three losses mentioned above:

[0077] (17).

[0078] A regional distributed photovoltaic power prediction system based on multi-data cross-modal fusion includes:

[0079] The data processing module is used for spatiotemporal data gridding consistency processing. It collects raw meteorological data and historical power data, performs data standardization, and then uses unified geographic coordinates as a benchmark to realize the spatial correlation between meteorological data and historical power data.

[0080] The extraction module adaptively extracts data based on the spatial features of AG-CNN with dynamic perceptual convolution. AG-CNN relies on a unified basic spatial benchmark and achieves local modeling through dynamic feature perception and cross-regional information fusion.

[0081] The design module is used for the design of cross-modal feature bridging module. Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, it transforms the high-dimensional local features output by AG-CNN into a spatiotemporal token sequence that can be parsed by DR-Transformer, while preserving key spatial details and cross-regional correlation information.

[0082] The prediction module is used to incorporate the DR-Transformer temporal correlation modeling with dual-dimensional dynamic attention. The DR-Transformer takes the spatial dimension + temporal dimension dynamic attention mechanism as its core, deeply integrates the long-term temporal dependence of historical power and the driving role of future meteorological data, realizes the accurate coupling of meteorological and power cross-modal spatiotemporal features, and obtains the total power prediction result of the distributed photovoltaic system.

[0083] The beneficial effects of this invention are as follows:

[0084] 1. This invention focuses on "precise coupling of cross-modal spatiotemporal features" and "deep capture of long-term time-series dependencies" as its core objectives. It relies on an implicit spatial benchmark (gridized meteorological-power correlation units) to associate meteorological data with photovoltaic locations. It combines AG-CNN's adaptive extraction capability of "local meteorological-power spatial features" with DR-Transformer's cross-modal fusion advantage of "long-term historical power data - future meteorological drivers" to deeply explore the dynamic coupling relationship between "meteorological disturbances and power fluctuations." This approach ensures spatial consistency of data through the implicit spatial benchmark while overcoming the limitations of fixed structures through dynamic mechanisms, adapting to the characteristics of distributed photovoltaic systems—"numerous points, wide distribution, and significant susceptibility to micro-meteorological disturbances"—and specifically addresses the challenges faced by existing models.

[0085] 2. The dynamic region modeling of AG-CNN in this invention provides a refined spatial foundation for global fusion. The cross-modal feature bridging module solves the heterogeneous adaptation problem between convolutional features and Transformer sequences, while DR-Transformer enables in-depth mining of cross-modal spatiotemporal correlations. The synergy of these three components allows the model to retain the power heterogeneity of different regions while accurately capturing the dynamic correlation between meteorological drivers and power time series, maintaining stable prediction performance even in complex meteorological scenarios such as cloudy and rainy weather. Attached Figure Description

[0086] Figure 1 This is a flowchart of the overall AG-CNN process of the present invention;

[0087] Figure 2 This is a schematic diagram of the DR-Transformer part of the present invention;

[0088] Figure 3 This is a diagram of the overall architecture of the DR-Transformer of the present invention;

[0089] Figure 4 This is a flowchart of the cross-modal feature bridging module (CMFB) of the present invention;

[0090] Figure 5 This is a flowchart of the method of the present invention;

[0091] Figure 6 This is a diagram showing the relative location distribution of photovoltaic power stations according to an embodiment of the present invention;

[0092] Figure 7 This is a comparison chart of power prediction results for different models in a photovoltaic power station according to an embodiment of the present invention. Figure 7 The middle (A) graph is a comparison of the power prediction results of different models in station A; Figure 7 The middle (B) graph is a comparison of the power prediction results of different models in station B; Figure 7 The middle (C) graph is a comparison of the power prediction results of different models in station C; Figure 7 The middle (D) graph is a comparison of the power prediction results of different models in station D; Figure 7 The middle (E) is a comparison chart of the power prediction results of different models in station E;

[0093] Figure 8 This is a comparison chart of error indices of different prediction models for various stations according to embodiments of the present invention, wherein, Figure 8 (A) is a comparison chart of error indices for model M1; Figure 8 (B) is a comparison chart of the error indices of model M2; Figure 8 The middle (C) section is a comparison chart of the error indices of model M3; Figure 8 The middle (D) graph is a comparison of the error indices of model M4;

[0094] Figure 9 Box plots comparing the prediction effects of different models at various stations according to embodiments of the present invention, wherein, Figure 9 (A) is a box plot comparing the prediction effects of different models for station A; Figure 9 (B) is a box plot comparing the prediction effects of different models for station B; Figure 9 The middle (C) is a box plot comparing the prediction effects of different models for station C; Figure 9 The middle (D) is a box plot comparing the prediction effects of different models for station D; Figure 9 The middle (E) is a box plot comparing the prediction effects of different models for station E. Detailed Implementation

[0095] The present invention will be further described below with reference to the embodiments and accompanying drawings, but is not limited thereto.

[0096] Example 1:

[0097] like Figures 1-5 As shown, this embodiment provides a regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion, the steps of which are as follows:

[0098] (1) Spatiotemporal data gridding consistency processing: collect the original meteorological data and historical power data, and standardize the data. Then, based on the unified geographic coordinates, realize the spatial correlation between meteorological data and historical power data.

[0099] Meteorological data includes solar irradiance, temperature, and humidity;

[0100] The data standardization process processes meteorological data and historical power data separately to eliminate the influence of different characteristic units. The Z-score standardization method is used, and the formula is as follows:

[0101] (1)

[0102] Where x represents the original data, μ represents the data mean, and σ represents the standard deviation. This is the standardized data.

[0103] Spatial correlation: Combining the coverage of photovoltaic arrays and meteorological monitoring points, and based on the spatial scale of meteorological data, the geographical location of photovoltaic panels is matched with the corresponding meteorological regions, so that power data and meteorological data form a gridded spatial correlation. At the same time, the data is time-aligned according to the time axis of prediction granularity, and abnormal data is removed, forming a spatiotemporally aligned input structure, which lays the foundation for subsequent models to capture spatiotemporal correlations.

[0104] (2) Based on the spatial features of AG-CNN with dynamic perception convolution, the data is adaptively extracted. AG-CNN relies on a unified basic spatial benchmark and achieves refined local modeling through dynamic feature perception and cross-regional information fusion. Its core is to build a feature extraction mechanism that has both spatial consistency and flexible adaptability. It can ensure the natural correlation between meteorological and power data in the spatial dimension and accurately capture the feature pattern differences in different regions.

[0105] The core design of AG-CNN is dynamic perceptive convolutional kernels and adaptive region fusion: using grid cells as the basic spatial index, the convolutional kernels of AG-CNN adapt in real time to the station distribution density and feature fluctuation patterns within different grid cells through learnable parameters. For example, in grid cells with dense stations, the kernel function automatically focuses on local details within the cell (subtle features such as irradiance gradient and cloud shadow edges); in grid cells with sparse stations, the kernel function flexibly extends to adjacent grid cells to capture a wider range of meteorological trends (such as cloud movement at the regional scale). This dynamic adjustment mechanism allows the model to naturally fit the actual spatial feature distribution and avoid mechanical dependence on fixed grid boundaries.

[0106] The convolutional kernel weights of AG-CNN are constructed collaboratively by a Gaussian function offset by spatial relative positions and learnable parameters.

[0107] (2)

[0108] Where (i, j) represents the spatial relative position offset, such as the offset of the surrounding area centered on the target area, i is the horizontal spatial relative position offset, j is the vertical spatial relative position offset, and σ g θ is the Gaussian kernel bandwidth, used to control the decay rate of the spatial influence range; s(i,j) is the feature similarity between the offset region and the target region, including information such as the consistency of power fluctuation patterns and irradiance change trends; θ is a learnable parameter, α=5, used to balance the influence of spatial location correlation and feature pattern similarity. When performing convolution operations on the feature map, the output features will adaptively fuse information from the target region and surrounding related regions.

[0109] (3)

[0110] Among them, y x，y To output the feature values ​​of the target region in the feature map, x x+i，y+j For features of surrounding related regions in the input feature map, such as irradiance, temperature, and average power within the region, the model can maintain natural spatial continuity and capture feature relationships between different regions through dynamic weights w(i,j), providing a feature foundation that combines spatial consistency and local detail for subsequent global modeling.

[0111] (3) Cross-modal feature bridging module design: Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, the high-dimensional local features output by AG-CNN are transformed into a spatiotemporal token sequence that can be parsed by DR-Transformer, while retaining key spatial details and cross-regional correlation information;

[0112] The Cross-Modal Feature Bridging Module (CMFB) adapts AG-CNN features to the input of the DR-Transformer. The feature map output by AG-CNN is first flattened into a one-dimensional sequence according to the region index, and then the channel dimension is mapped to the hidden dimension of the DR-Transformer through linear projection to form the initial token sequence. To balance the preservation of local features with the adaptability of global modeling, CMFB dynamically adjusts the feature weights through a gating mechanism: for regions with high photovoltaic capacity and large power fluctuations, the feature weights are strengthened; for regions without sites or with stable features, noise interference is weakened. At the same time, the temporal features of historical power are introduced, and the spatial features and temporal context are explicitly fused through the Spatiotemporal Gating Unit (STGU) to generate a token sequence that carries both regional details and temporal correlation, ensuring a smooth transition of features from the convolutional feature map to the Transformer sequence.

[0113] The feature map F output by AG-CNN is a four-dimensional tensor:

[0114] (4)

[0115] Where B is the training batch size, C is the number of feature channels, and H×W corresponds to the number of region divisions;

[0116] First, the two-dimensional region structure is transformed into a one-dimensional sequence through spatial flattening, and the channel dimension is adapted to the hidden dimension of the Transformer through learnable linear transformation:

[0117] (5)

[0118] in, The flattened sequence features are represented by N = H × W, where N is the total number of regions. Let be the projection matrix. Here, D is the bias term and D is the hidden dimension of the Transformer. This step compresses the local features of each region into tokens of fixed dimensions, forming a sequence of length N, which provides the basic unit for global modeling.

[0119] Since the optimization objectives of AG-CNN and DR-Transformer differ, a reparameterization layer is introduced to dynamically adjust the feature distribution to alleviate training bias:

[0120] (6)

[0121] in, For element-wise gating coefficients, It is the Sigmoid activation function. , As learnable parameters, the gating mechanism achieves dynamic alignment of feature distributions between the two models by strengthening key features and suppressing noise;

[0122] To display the temporal features of fused historical power and the spatial features of AG-CNN, a spatiotemporal gating unit (STGU) is introduced, and the correlation weights between the two are dynamically calibrated through cross-modal attention:

[0123] Let the historical power sequence be , among which, T h Let R be the historical time step, B be the training batch size, N be the total number of regions, and D be the Transformer hidden dimension. Then the gated fusion process is as follows:

[0124] (7)

[0125] in, F is the dynamic fusion coefficient. in The output features of the Space-Time Gated Unit (STGU) carry both spatial details and temporal correlations. The STGU is a space-time gated unit used to fuse spatial and temporal features. α For the cross-modal attention weight matrix, b α For cross-modal attention bias term, F reparam The reparameterized spatial features output by equation (6), As a cross-modal attention mechanism, it achieves semantic-level calibration of spatial and temporal features by learning the association weights between them.

[0126] Final output Directly used as input to the DR-Transformer encoder, each token retains both the local spatial details extracted by AG-CNN and incorporates the temporal context of historical power, providing a high-quality feature foundation for the subsequent dynamic attention mechanism to accurately capture the dual dependence of "spatial correlation-temporal evolution".

[0127] (4) DR-Transformer, as the core of global spatiotemporal fusion, realizes cross-modal correlation modeling of historical power and future weather through encoder-decoder structure;

[0128] The encoder receives the sequence output from the cross-modal feature bridging module, processes historical power and spatial features, and captures implicit correlations and long-term temporal dependencies between regions through dynamic spatial attention. It employs a multi-layered stacked structure: the bottom layer introduces a dynamic window partitioning strategy, strengthening the temporal correlation of local regions through attention calculation within the window; the upper layer uses global attention to capture long-distance temporal dependencies. Each layer contains multi-head dynamic spatial attention and a feedforward neural network, where the attention weights are modulated by both the relative position of the regions and the similarity of power features. For example, for regions with high historical power correlation coefficients, the attention ratio between them is automatically increased.

[0129] The decoder processes meteorological data for future time steps. Under the constraint of the masked self-attention mechanism, it dynamically couples meteorological features with historical features output by the encoder through cross-attention. It adopts a multi-layer stacked structure, with each layer consisting of multi-head masked self-attention, multi-head cross-attention, and a feedforward neural network: masked self-attention ensures the temporal causality of the meteorological sequence; cross-attention adjusts the weights through dynamic correlation coefficients, for example, strengthening the attention ratio of irradiance during noon and increasing the driving weight of cloud cover during cloudy periods. After mapping the encoder and decoder features to the same dimension through a linear projection layer, cross-attention achieves a precise correlation between historical power trends and future meteorological drivers, outputting features that fuse global spatiotemporal information.

[0130] Finally, the output layer converts the fused features into regional power prediction values ​​for multiple future time steps through a fully connected layer. Then, based on the photovoltaic installed capacity of each region, the values ​​are weighted and summed to obtain the total power prediction result of the distributed photovoltaic system, completing the entire process from input data to prediction output.

[0131] The core design of DR-Transformer focuses on a dynamic attention mechanism in both spatial and temporal dimensions: In the spatial dimension, attention weights are adaptively adjusted based on the similarity of regional features and their relative positional relationships to avoid mechanical dependence on fixed spatial units; in the temporal dimension, the evolution of historical power and the driving force of future weather are dynamically coupled through a cross-attention mechanism to achieve deep fusion of cross-modal features. Its design point is not about following a fixed structure, but about how to make the correlation modeling of spatial and temporal features more in line with the actual fluctuation patterns of distributed photovoltaics through the dynamic adjustment of the attention mechanism.

[0132] To address the discrete distribution characteristics of distributed photovoltaic (PV) sites, the DR-Transformer encoder captures implicit correlations between regions through dynamic spatial attention. The attention weight calculation considers both relative spatial distance and feature pattern similarity. The formula for calculating the dynamic spatial attention weight is as follows:

[0133] (8)

[0134] Where q, k, and v are the query, key, and value vectors, respectively, and d k For feature dimension, β(r) is the feature similarity coefficient (calculated based on the cosine similarity of the regional power sequence, with a value of 0-1), and β(r) is the relative spatial distance coefficient, which is generated based on the Gaussian decay function and reflects the spatial correlation between regions. Through the coordinated modulation of the two, the attention weight can not only reflect the natural continuity in space, but also focus on the related regions with similar features, avoiding the correlation fragmentation caused by fixed spatial division.

[0135] The DR-Transformer's decoder dynamically fuses future meteorological data with historical features from the encoder output, achieving cross-modal correlation through masked self-attention and cross-attention.

[0136] Future meteorological data has a natural time-series dependency, but directly using self-attention may lead to future information being focused on. Masked self-attention forces the model to follow temporal causality through the lower triangular matrix, allowing only the current time step to focus on historical and concurrent meteorological characteristics.

[0137] Let the future weather sequence be the input to the decoder. T represents the number of future prediction steps, D represents the dimension of the meteorological feature vector, R represents the real number field, and m t Given the meteorological feature vector at step t, a query is generated through linear projection. ,key ,value , , , , where is the meteorological feature projection matrix, and h is the number of attention heads;

[0138] For time step t, the similarity of the mask self-attention is calculated as follows:

[0139] (9)

[0140] in, For the time step that is being noticed, This is a scaling factor to prevent the softmax gradient from vanishing due to excessively large inner product values.

[0141] Introducing a lower triangular mask matrix Suppressing future information through masking:

[0142] (10)

[0143] Final The masked self-attention output of the step is:

[0144] (11)

[0145] in, To output the projection matrix, the multi-head attention results are concatenated and projected to obtain meteorological features with temporal dependencies. .

[0146] In distributed photovoltaic data, the contribution of different regions to the total power varies significantly, and power prediction needs to meet the requirements of overall fitting accuracy and spatiotemporal correlation. A single loss function is difficult to satisfy these requirements. Therefore, a multi-objective loss function is designed to collaboratively optimize the model parameters, which not only retains the role of spatial benchmark in integrating discrete data, but also enhances the model's ability to capture dynamic weather-power correlation.

[0147] As the core foundation of the loss function, it is necessary to prioritize ensuring the model's fitting accuracy to the overall power level. Considering that photovoltaic power prediction is a continuous value regression problem, and that the penalty for larger errors is more stringent in practical applications, the squared error penalty characteristic of MSE can amplify the impact of extreme biases, forcing the model to focus on reducing large power errors. At the same time, its continuous differentiability is compatible with the model's gradient descent optimization mechanism, ensuring the stability of the training process. Therefore, the method constructs a multi-objective loss function oriented towards regional characteristics and cross-modal constraints:

[0148] The mean squared error (MSE) is used as the basic loss term, and the formula is as follows:

[0149] (12)

[0150] Where N is the total number of regions. To predict the time step, and Let be the predicted power and the actual power of the i-th region at time t, respectively;

[0151] AG-CNN extracts local features through dynamic kernels and cross-regional fusion, but its output regional features may suffer from fragmented correlations between adjacent regions due to the randomness of dynamic neighborhoods. The core of the regional feature consistency loss is to force the model to respect the implicit correlations between regions in physical space, ensuring that feature extraction not only focuses on the local details of a single region, but also captures the collaborative changes in adjacent regions caused by common meteorological disturbances.

[0152] (13)

[0153] Where, N i For the dynamic neighborhood of region i, w i，j The feature similarity weights for regions i and j are as follows: the higher the similarity, the greater the weight. This term compensates for the spatial correlation fragmentation problem caused by fixed grid division by constraining the consistency of power differences between adjacent regions.

[0154] The core of DR-Transformer is to fuse historical power time-series features with future meteorological driving features through cross-attention. However, the model may exhibit prediction bias due to an unreasonable allocation of attention weights, resulting in a "disconnect between meteorological and power trends." The goal of cross-modal time-series correlation loss is to force the model to learn the physical causal relationship between meteorological factors and power changes, ensuring that cross-modal correlations in the time-series dimension conform to objective laws, rather than merely fitting the surface correlation of data. It regulates the model's correlation learning from the physical mechanism level through dual constraints: trend consistency + reasonable attention weights.

[0155] (14)

[0156] in, As a trend consistency loss function, it is used to constrain the physical rationality of the direction of change. β is the weighting coefficient of the trend consistency loss, used to balance the strength of trend constraint and attention weight constraint.

[0157] (15)

[0158] in, To predict the time-series variation of power, This represents the temporal variation of meteorological factors. For cross-entropy loss, a high penalty is imposed for situations where the power trend is opposite to the meteorological trend;

[0159] As a cross-modal attention loss function, it is used to constrain the objective correlation of attention weights:

[0160] (16)

[0161] in, In the DR-Transformer cross-attention, meteorological features affect the region. exist Attention weight at any moment This is the true correlation coefficient between meteorology and power calculated based on historical data.

[0162] The overall loss function is the weighted sum of the three losses mentioned above:

[0163] (17)

[0164] Example Analysis:

[0165] This embodiment features targeted configuration of manual parameters. AG-CNN employs a 3-layer convolutional block cascade structure. The first layer captures instantaneous features within a region using a dynamic kernel; the second layer extracts cross-regional meteorological propagation patterns using a dynamic kernel, outputting 64 feature channels; the third layer fuses the features from the first two layers through convolution, outputting 128 feature channels. Each convolutional layer is processed with batch normalization and the LeakyReLU activation function (slope set to 0.01). The DR-Transformer's encoder and decoder both use a multi-layer stacked structure, with a hidden dimension of 128, an 8-head multi-head self-attention mechanism, and a feedforward neural network dimension of 512. The encoder introduces a dynamic window partitioning strategy with a 4×4 window size; the linear transformation matrix for cross-attention in the decoder has a dimension of 128×128. In the Cross-Modal Feature Bridging Module (CMFB), the adaptive projection matrix has a dimension of 128×128, and the initial value of the bias term is 0.01; the cross-modal attention weight matrix of the Spatiotemporal Gated Unit (STGU) has a dimension of 128×128, the temporal feature embedding dimension is 64, and the loss function parameters are as follows. =0.3, =0.5, =0.1.

[0166] 3.1 Dataset and Experimental Environment Configuration;

[0167] To verify the advantages of the proposed AG-CNN and DR-Transformer combined model in "power-meteorological cross-modal deep fusion" and "accurate modeling of long time-series dependencies", five distributed photovoltaic power stations (AEs) in a certain region were selected, and the relative locations of the stations are shown in the figure. Figure 6 Experiments were conducted using actual data. The dataset covers all 15-minute data points for a given year, including power generation data from various power plants and meteorological data (solar irradiance, cloud cover, temperature, etc.) for the corresponding regions. The data was Z-score normalized and spatiotemporally aligned.

[0168] The experiment was implemented using Python 3.9, with a hardware configuration of one GTX 2060 graphics card, 16GB of memory, and a 6-core Intel(R) Core (TM) i7-10750H CPU @ 2.60GHz.

[0169] To systematically verify the comprehensive performance of this embodiment (M4) in distributed photovoltaic power prediction tasks, three representative prediction models were selected as a benchmark for comparison. The BP neural network (M1), as a classic feedforward neural network model, represents a basic power prediction method; the TCN-Transformer model (M2), combining a temporal convolutional network and the Transformer architecture, is a commonly used model in long-sequence prediction; and the GCN-LSTM model (M3), through the combination of a graph convolutional network and a long short-term memory network, can simultaneously handle spatial and temporal dependencies. Through comparative analysis with these three models, the predictive ability and adaptability of the proposed model in complex meteorological scenarios can be evaluated from multiple dimensions, and its performance in power-meteorological cross-modal fusion and long-sequence dependency modeling can be systematically examined.

[0170] 3.2 Predictive performance comparison and error analysis;

[0171] To visually compare the differences in power fitting performance of the four models in real-world application scenarios, we selected five random days of measured data covering normal and fluctuating scenarios from five distributed photovoltaic power stations of AE. We then conducted a comparative analysis of the prediction results of the M1 (BP neural network), M2 (TCN-Transformer), M3 (GCN-LSTM), and the M4 (AG-CNN+DR-Transformer) model proposed in this embodiment.

[0172] based on Figure 7 Comparative analysis of the power prediction curves from various power plants reveals that the M4 model (AG-CNN+DR-Transformer) proposed in this embodiment exhibits significant advantages in curve fitting accuracy, dynamic response speed, and adaptability to complex scenarios. Its prediction curves closely track actual power changes, especially during the midday period in fluctuating scenarios. The M4 model not only accurately captures the fluctuation amplitude but also achieves precise temporal synchronization. In contrast, the M1 model (BP neural network) shows a significant smoothing bias at power abrupt changes, the M2 model (TCN-Transformer), while able to grasp the trend, suffers from phase lag, and the M3 model (GCN-LSTM) exhibits overshoot in peak tracking. This visualization confirms the effectiveness of the AG-CNN module in enhancing local feature extraction through dynamic convolutional kernels and the DR-Transformer module in deepening temporal correlation through cross-modal attention, jointly ensuring the model's prediction robustness across multiple distributed photovoltaic scenarios.

[0173] To more intuitively quantify and analyze the prediction performance of the proposed model, this paper uses two indicators, normalized root mean square error (NRMSE) and normalized mean absolute error (NMAE), to evaluate the prediction accuracy of the model.

[0174] (18)

[0175] (19)

[0176] Where n is the total number of samples. and P represents the actual power and the predicted power of the j-th sample, respectively. N The rated power is used as the reference for normalization.

[0177] Table 1: Error Indicators of Each Prediction Model

[0178]

[0179] Table 1 quantifies the differences in prediction accuracy of the four models across five sites under normal and fluctuating scenarios using two core indicators: Normalized Root Mean Square Error (NRMSE) and Normalized Mean Absolute Error (NMAE). Figure 8 The numerical results are visualized in the form of bar charts, and the two mutually corroborate each other, highlighting the significant advantages of the M4 model in this paper. In normal scenarios, the NRMSE and NMAE of M4 are as low as 0.0821-0.1069 and 0.0697-0.1241, respectively, which are 35%-40% lower than M1 on average, 18%-25% lower than M2 on average, and 10%-15% lower than M3 on average. In fluctuating scenarios, M4 has the smallest error increase, with NRMSE and NMAE remaining at 0.1006-0.1269 and 0.0897-0.1341, respectively, still significantly better than M1 which relies on static structures, M2 which fuses shallow layers across modalities, and M3 which is limited by fixed graph structures and convolutional kernels.

[0180] Figure 8 Further analysis revealed the spatial distribution and stability differences of errors in each model. The M4 histogram was generally low and fluctuated gently, with NRMSE and NMAE fluctuations of only 0.0248 and 0.0544 for all stations. Even at sparsely populated stations D and E or frequently disturbed meteorological station C, it maintained a low error level, demonstrating the synergistic effectiveness of AG-CNN dynamic spatial modeling and DR-Transformer cross-modal long-term fusion. In contrast, the M1 histogram was generally high and fluctuated the most, while M2 and M3 showed significant error peaks in areas with long-term meteorological changes or sparse stations. Their limitations contrasted sharply with the innovative mechanism of M4, confirming M4's core advantages in adapting to the spatial heterogeneity of distributed photovoltaic systems and complex meteorological scenarios.

[0181] To further illustrate the prediction performance of this embodiment, the prediction results of the proposed model and the comparative model are compared and analyzed in different ways.

[0182] Figure 9 The distribution characteristics of the absolute prediction errors of each model are clearly presented: M1 (BP neural network) has a wide box and dense extreme values. Due to regional distribution differences and insufficient modeling of modal correlation, the error dispersion is extremely high. Although the box of M2 (TCN-Transformer) is narrower than that of M1, there are still many outliers, reflecting that its shallow cross-modal fusion leads to limited stability. M3 (GCN-LSTM) has a low median. The combination of spatial and temporal modeling alleviates some error fluctuations, but the fixed structure still leaves room for optimization in dynamic scenes. M4 (AG-CNN+DR-Transformer) performs the best. The box is extremely narrow and there are no extreme outliers. Its dynamic spatial feature extraction of AG-CNN and cross-modal long temporal fusion of DR-Transformer work together to effectively suppress error dispersion and maintain stable performance in different regions.

[0183] Example 2:

[0184] This embodiment provides a regional distributed photovoltaic power prediction system based on multi-data cross-modal fusion, including:

[0185] The data processing module is used for spatiotemporal data gridding consistency processing. It collects raw meteorological data and historical power data, performs data standardization, and then uses unified geographic coordinates as a benchmark to realize the spatial correlation between meteorological data and historical power data.

[0186] The extraction module adaptively extracts data based on the spatial features of AG-CNN with dynamic perceptual convolution. AG-CNN relies on a unified basic spatial benchmark and achieves local modeling through dynamic feature perception and cross-regional information fusion.

[0187] The design module is used for the design of cross-modal feature bridging module. Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, it transforms the high-dimensional local features output by AG-CNN into a spatiotemporal token sequence that can be parsed by DR-Transformer, while preserving key spatial details and cross-regional correlation information.

[0188] The prediction module is used to incorporate the DR-Transformer temporal correlation modeling with dual-dimensional dynamic attention. The DR-Transformer takes the spatial dimension + temporal dimension dynamic attention mechanism as its core, deeply integrates the long-term temporal dependence of historical power and the driving role of future meteorological data, realizes the accurate coupling of meteorological and power cross-modal spatiotemporal features, and obtains the total power prediction result of the distributed photovoltaic system.

Claims

1. A regional distributed photovoltaic power prediction method based on multi-source data cross-modal fusion, characterized in that, The steps are as follows: (1) Spatiotemporal data gridding consistency processing: collect the original meteorological data and historical power data, and standardize the data. Then, based on the unified geographic coordinates, realize the spatial correlation between meteorological data and historical power data. Spatial correlation: Combining the coverage of photovoltaic arrays and meteorological monitoring points, based on the spatial scale of meteorological data, the geographical location of photovoltaic panels is matched with the corresponding meteorological area, so that power data and meteorological data form a gridded spatial correlation. At the same time, the data is time-aligned according to the time axis of prediction granularity, and abnormal data is removed to form a spatiotemporally aligned input structure. (2) Based on the spatial features of AG-CNN with dynamic perception convolution, the data is adaptively extracted. AG-CNN relies on a unified basic spatial benchmark and achieves local modeling through dynamic feature perception and cross-regional information fusion. Using grid cells as the basic spatial index, the convolutional kernels of AG-CNN adapt in real time to the site distribution density and feature fluctuation patterns within different grid cells through learnable parameters; The convolutional kernel weights of AG-CNN are constructed collaboratively by a Gaussian function offset by spatial relative positions and learnable parameters. （2） Where (i, j) represents the spatial relative position offset, i is the horizontal spatial relative position offset, j is the vertical spatial relative position offset, and σ g Let be the Gaussian kernel bandwidth, s(i,j) be the feature similarity between the offset region and the target region, θ be a learnable parameter, and α=5 be used to balance the influence of spatial location correlation and feature pattern similarity. When performing convolution operations on the feature map, the output features adaptively fuse information from the target region and surrounding related regions. （3） Among them, y x，y To output the feature values ​​of the target region in the feature map, x x+i，y+j The input feature map contains features of the surrounding associated regions; (3) Cross-modal feature bridging module design: Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, the high-dimensional local features output by AG-CNN are transformed into a spatiotemporal token sequence that can be parsed by DR-Transformer, while retaining key spatial details and cross-regional correlation information; (4) Integrating dual-dimensional dynamic attention into DR-Transformer temporal correlation modeling, DR-Transformer takes spatial dimension + temporal dimension dynamic attention mechanism as the core, deeply integrates the long-term temporal dependence of historical power and the driving role of future meteorological data, realizes the accurate coupling of meteorological-power cross-modal spatiotemporal characteristics, and obtains the total power prediction result of distributed photovoltaic system; DR-Transformer achieves cross-modal correlation modeling of historical power and future weather through an encoder-decoder structure; The encoder receives the sequence output by the cross-modal feature bridging module, processes historical power and spatial features, and captures implicit associations and long-term temporal dependencies between regions through dynamic spatial attention. It adopts a multi-layer stacked structure: the bottom layer introduces a dynamic window partitioning strategy to strengthen the temporal association of local regions through attention calculation within the window; the upper layer uses global attention to capture long-distance temporal dependencies. Each layer contains multi-head dynamic spatial attention and a feedforward neural network, where the attention weights are modulated by both the relative position of the regions and the similarity of power features. The decoder processes meteorological data for future time steps. Under the constraint of the masked self-attention mechanism, it dynamically couples meteorological features with historical features output by the encoder through cross-attention. It adopts a multi-layer stacked structure, with each layer consisting of multi-head masked self-attention, multi-head cross-attention, and a feedforward neural network: masked self-attention ensures the temporal causality of the meteorological sequence; cross-attention adjusts the weights through dynamic correlation coefficients. After mapping the encoder and decoder features to the same dimension through a linear projection layer, cross-attention achieves a precise correlation between historical power trends and future meteorological drivers, outputting features that fuse global spatiotemporal information. Finally, the output layer converts the fused features into regional power prediction values ​​for multiple future time steps through a fully connected layer. Then, based on the photovoltaic installed capacity of each region, the values ​​are weighted and summed to obtain the total power prediction result of the distributed photovoltaic system, completing the entire process from input data to prediction output.

2. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 1, characterized in that, In step (1), the meteorological data includes solar irradiance, temperature, and humidity; The data standardization process processes meteorological data and historical power data separately to eliminate the influence of different characteristic units. The Z-score standardization method is used, and the formula is as follows: （1） Where x represents the original data, μ represents the data mean, and σ represents the standard deviation. This is the standardized data.

3. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 2, characterized in that, The AG-CNN network structure adopts a multi-layer convolutional block serial design: the first layer captures instantaneous irradiance changes in the region through dynamic kernels; the second layer extracts the meteorological propagation patterns across regions through dynamic kernels; the third layer fuses the features of the first two layers through convolution to enhance key details, and each convolution is followed by batch normalization and activation processing.

4. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 3, characterized in that, In step (3), the feature map F output by AG-CNN is a four-dimensional tensor: （4） Where B is the training batch size, C is the number of feature channels, and H×W corresponds to the number of region divisions; First, the two-dimensional region structure is transformed into a one-dimensional sequence through spatial flattening, and the channel dimension is adapted to the hidden dimension of the Transformer through learnable linear transformation: （5） in, The flattened sequence features are represented by N = H × W, where N is the total number of regions. Let be the projection matrix. Here, D is the bias term, and D is the hidden dimension of the Transformer. Since the optimization objectives of AG-CNN and DR-Transformer differ, a reparameterization layer is introduced to dynamically adjust the feature distribution to alleviate training bias: （6） in, For element-wise gating coefficients, It is the Sigmoid activation function. , As learnable parameters, the gating mechanism achieves dynamic alignment of feature distributions between the two models by strengthening key features and suppressing noise; To display the temporal features of fused historical power and the spatial features of AG-CNN, a spatiotemporal gating unit is introduced, and the correlation weights between the two are dynamically calibrated through cross-modal attention: Let the historical power sequence be , among which, T h Let R be the historical time step, B be the training batch size, N be the total number of regions, and D be the Transformer hidden dimension. Then the gated fusion process is as follows: （7） in, F is the dynamic fusion coefficient. in The output features of STGU carry both spatial details and temporal correlations. STGU is a spatiotemporal gating unit used to fuse spatial and temporal features. α For the cross-modal attention weight matrix, b α For cross-modal attention bias term, F reparam The reparameterized spatial features output by equation (6), As a cross-modal attention mechanism, it achieves semantic-level calibration of spatial features and temporal features by learning the association weights between them. Final output It is used directly as the input to the DR-Transformer encoder.

5. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 4, characterized in that, DR-Transformer focuses on a dynamic attention mechanism in both spatial and temporal dimensions: in the spatial dimension, it adaptively adjusts attention weights based on the similarity of regional features and their relative positional relationships. In the temporal dimension, the evolution of historical power and the driving role of future weather are dynamically coupled through a cross-attention mechanism; To address the discrete distribution characteristics of distributed photovoltaic (PV) sites, the DR-Transformer encoder captures implicit correlations between regions through dynamic spatial attention. The attention weight calculation considers both relative spatial distance and feature pattern similarity. The formula for calculating the dynamic spatial attention weight is as follows: （8） Where q, k, and v are the query, key, and value vectors, respectively, and d k For feature dimension, β(r) is the feature similarity coefficient, and β(r) is the relative spatial distance coefficient. Through the coordinated modulation of the two, the attention weight can reflect the natural continuity in space and also focus on the related regions with similar features. The DR-Transformer's decoder dynamically fuses future meteorological data with historical features from the encoder output, achieving cross-modal correlation through masked self-attention and cross-attention. Masked self-attention forces the model to follow temporal causality through a lower triangular matrix, allowing the current time step to focus only on historical and concurrent meteorological features; Let the future weather sequence be the input to the decoder. T represents the number of future prediction steps, D represents the dimension of the meteorological feature vector, R represents the real number field, and m t Given the meteorological feature vector at step t, a query is generated through linear projection. ,key ,value , , , , where is the meteorological feature projection matrix, and h is the number of attention heads; For time step t, the similarity of the mask self-attention is calculated as follows: （9） in, For the time step that is being noticed, This is the scaling factor; Introducing a lower triangular mask matrix Suppressing future information through masking: （10） In the formula, This represents the t-th row and the th element in the mask matrix. The elements of the column, MaskedSim is the attention similarity score after masking; Final The masked self-attention output MaskedAttn(t) of the step is: （11） in, The output projection matrix is ​​given by v, which represents the value vector. softmax(·) is the normalization exponential function, and Concat(·) represents the multi-head attention concatenation operation. After concatenation and projection, the meteorological features containing time-dependent characteristics are obtained from the multi-head attention results. .

6. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 5, characterized in that, The method constructs a multi-objective loss function that is oriented towards regional characteristics and cross-modal constraints.

7. The regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 6, characterized in that, The multi-objective loss function uses the mean squared error as the basic loss term, and the formula is as follows: （12） Where N is the total number of regions. To predict the time step, and Let be the predicted power and the actual power of the i-th region at time t, respectively; AG-CNN extracts local features through dynamic kernels and cross-regional fusion, but the randomness of its output regional features leads to a break in the correlation between adjacent regions due to the randomness of the dynamic neighborhood. The core of the regional feature consistency loss is to force the model to respect the implicit correlation between regions in the physical space, ensuring that feature extraction not only focuses on the local details of a single region, but also captures the collaborative changes in adjacent regions caused by common meteorological disturbances. （13） Where, N i For the dynamic neighborhood of region i, w i，j The feature similarity weights for regions i and j are given; the higher the similarity, the greater the weight. The core of DR-Transformer is to fuse historical power time-series features with future meteorological driving features through cross-attention. The goal of the cross-modal time-series association loss is to force the model to learn the physical causal relationship between meteorological factors and power changes, ensuring that the cross-modal association in the time-series dimension conforms to objective laws, rather than merely fitting the surface correlation of the data. It regulates the model's association learning from the physical mechanism level through dual constraints: （14） in, As a trend consistency loss function, it is used to constrain the physical rationality of the direction of change. β is the weighting coefficient of the trend consistency loss, used to balance the strength of trend constraint and attention weight constraint. （15） in, To predict the time-series variation of power, This represents the temporal variation of meteorological factors. For cross-entropy loss, a high penalty is imposed for situations where the power trend is opposite to the meteorological trend; As a cross-modal attention loss function, it is used to constrain the objective correlation of attention weights: （16） in, In the DR-Transformer cross-attention, meteorological features affect the region. exist Attention weight at any moment The actual correlation coefficient between meteorology and power is calculated based on historical data; The overall loss function is the weighted sum of the three losses mentioned above: （17）。 8. A regional distributed photovoltaic power prediction system based on multi-data cross-modal fusion, applied to the regional distributed photovoltaic power prediction method based on multi-data cross-modal fusion as described in claim 1, characterized in that, include: The data processing module is used for spatiotemporal data gridding consistency processing. It collects raw meteorological data and historical power data, performs data standardization, and then uses unified geographic coordinates as a benchmark to realize the spatial correlation between meteorological data and historical power data. The extraction module adaptively extracts data based on the spatial features of AG-CNN with dynamic perceptual convolution. AG-CNN relies on a unified basic spatial benchmark and achieves local modeling through dynamic feature perception and cross-regional information fusion. The design module is used for the design of cross-modal feature bridging module. Through a three-level processing flow of adaptive projection-feature reparameterization-spatiotemporal fusion gating, it transforms the high-dimensional local features output by AG-CNN into a spatiotemporal token sequence that can be parsed by DR-Transformer, while preserving key spatial details and cross-regional correlation information. The prediction module is used to incorporate the DR-Transformer temporal correlation modeling with dual-dimensional dynamic attention. The DR-Transformer takes the spatial dimension + temporal dimension dynamic attention mechanism as its core, deeply integrates the long-term temporal dependence of historical power and the driving role of future meteorological data, realizes the accurate coupling of meteorological and power cross-modal spatiotemporal features, and obtains the total power prediction result of the distributed photovoltaic system.

Images