A data intelligent processing method and system for green electricity

Abstract

This invention discloses a method and system for intelligent data processing for green electricity, relating to the field of power grid management technology. The method includes constructing a green electricity generation prediction model and training it using historical multi-source data. The trained green electricity generation prediction model predicts future power generation. A reinforcement learning algorithm is used to define the state space, action space, and reward function, and to construct a green electricity supply and demand balance model. Based on future power generation and user electricity demand, the optimal supply and demand balance is obtained, and an optimal scheduling strategy is generated based on the supply and demand balance. The invention uses a temporal convolutional network (TCN) to construct and train the green electricity generation prediction model, effectively improving the accuracy and robustness of future power generation predictions.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid management, and in particular to a method and system for intelligent data processing of green electricity. Background Art

[0002] With the transformation of the global energy structure and the advancement of sustainable development goals, the efficient utilization and intelligent scheduling of green electricity (green power), a clean energy source, have become research hotspots in the energy sector. Green power generation is affected by multiple factors, such as meteorological conditions and equipment operating status, and is subject to volatility and uncertainty, posing significant challenges to balancing supply and demand in power grids. In recent years, data-driven artificial intelligence technologies have been widely applied in the green power sector. Existing prediction models based on time series analysis, convolutional neural networks (CNNs), and long-short-term memory networks (LSTMs) have improved the accuracy of green power generation forecasts to a certain extent, but they still lack the ability to integrate multi-source data and model long-term dependencies. Furthermore, existing scheduling strategies often rely on static rules or simple optimization algorithms, lacking the ability to respond in real time to dynamic changes in supply and demand, making it difficult to achieve efficient green power utilization and stable power grid operation.

[0003] The main shortcomings of existing technologies are reflected in the following two aspects: First, the ability to integrate multi-source data is limited. Existing methods usually process meteorological data, equipment operation data, and electricity consumption data independently, lacking an effective feature fusion mechanism, which makes it difficult for the model to fully capture the complex correlations between multi-source data. Secondly, the supply and demand balance model is not dynamic and intelligent enough. Existing scheduling strategies are mostly based on fixed rules or local optimization algorithms, which are difficult to adapt to the dynamic changes in green electricity generation and electricity demand, and cannot achieve the global optimal supply and demand balance. These problems limit the efficient use of green electricity and the stable operation of the power grid. There is an urgent need for an intelligent processing method that can deeply integrate multi-source data and dynamically optimize the supply and demand balance. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a data intelligent processing method for green electricity to solve the problems of insufficient multi-source data fusion and insufficient dynamics of the supply and demand balance model.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for intelligent data processing of green electricity, which includes collecting multi-source data and preprocessing the multi-source data, extracting multi-source features based on the multi-source data, and generating a comprehensive feature vector through a graph attention fusion method, wherein the multi-source data includes real-time meteorological data, equipment operation data and historical electricity consumption data; constructing a green electricity generation prediction model and using historical multi-source data for training, and predicting the power generation in a future period through the trained green electricity generation prediction model; using a reinforcement learning algorithm to define the state space, action space and reward function, and constructing a green electricity supply and demand balance model, obtaining the optimal value of the supply and demand balance based on the power generation and user electricity demand in the future period, and generating an optimal scheduling strategy based on the optimal value of the supply and demand balance.

[0008] As a preferred solution of the data intelligent processing method for green electricity described in the present invention, the preprocessing includes data cleaning, normalization, missing value processing, data alignment and denoising.

[0009] As a preferred solution of the data intelligent processing method for green electricity of the present invention, wherein: the method of extracting multi-source features based on multi-source data and generating a comprehensive feature vector by graph attention fusion method is as follows:

[0010] Based on real-time meteorological data, meteorological features are extracted through wavelet transform and principal component analysis;

[0011] Based on the equipment operation data, the equipment operation characteristics are extracted through time series analysis and convolutional neural network;

[0012] Based on historical electricity consumption data, electricity consumption characteristics are extracted through Fourier transform;

[0013] Multi-source features are input into the support vector machine, and meteorological features, equipment operation features, and electricity consumption features are mapped into a high-dimensional space using the RBF kernel to identify the correlation between features.

[0014] A two-level graph attention mechanism consisting of local and global levels is used to construct multi-source features into graph-structured data, where each multi-source feature is regarded as a node and the correlation between multi-source features is regarded as an edge.

[0015] Based on graph structure data, at the local level, a single-layer GAT is used to learn the relationship between nodes and directly adjacent nodes, and at the global level, a cross-layer attention mechanism is used to learn the relationship between nodes and indirectly adjacent nodes;

[0016] Based on the relationship between the node and its direct and indirect adjacent nodes, the attention weights at different levels are fused through the multi-head attention mechanism to generate a comprehensive feature vector F, which is expressed as:

[0017]

[0018] Among them, K is the number of heads of the multi-head attention mechanism, Q k is the query vector of the kth attention head, K k is the key vector of the kth attention head, V k is the value matrix of the kth attention head, d k is the dimension of multi-source features, and T is the transpose operator.

[0019] As a preferred solution of the data intelligent processing method for green electricity of the present invention, the green electricity generation prediction model is constructed and trained using historical multi-source data, and the specific steps are as follows:

[0020] Initialize the number of TCN layers, number of neurons, and learning rate based on the statistical characteristics and distribution of historical green power generation;

[0021] Construct an input layer based on the dimensions of the input data;

[0022] Construct hidden layers based on the initialized number of TCN layers, number of neurons, and learning rate to capture long-term dependencies in the feature vectors in time series format;

[0023] Use activation functions to construct a fully connected layer to output the green electricity generation amount with consistent dimensions;

[0024] Build a green electricity generation prediction model based on the input layer, hidden layer and fully connected layer;

[0025] The green electricity generation prediction model is trained using historical multi-source data, and the mean square error is used as the loss function. Backpropagation is used to update the parameters of the green electricity generation prediction model, and the Adam optimizer is used to dynamically adjust the learning rate.

[0026] As a preferred solution of the method for intelligent data processing of green electricity according to the present invention, the following specific steps are used to generate a prediction model for green electricity after training to predict power generation in the future:

[0027] Divide the comprehensive feature vector F into time steps, generate feature vectors in time series format, and perform normalization;

[0028] The normalized time series format feature vector is input into the green electricity generation prediction model to predict the power generation in the future. The expression is:

[0029] Y t+1:t+T =ReLU(TCN(X t-L+1:t ;W1,b1))·W2+b2;

[0030] Among them, Y t+1:t+Trepresents the power generation from time step t+1 to time step t+T, TCN represents the temporal convolutional network, X t-L+1:t It represents the normalized feature vector in the time series format from time step t-L+1 to time step t, W1 represents the weight of the TCN layer, b1 represents the bias parameter of the TCN layer, W2 is the weight of the fully connected layer, b2 represents the bias parameter of the fully connected layer, T represents the length of the future time period, and t represents the current time.

[0031] As a preferred solution of the data intelligent processing method for green electricity of the present invention, wherein: the reinforcement learning algorithm is used to define the state space, action space and reward function, and to construct the green electricity supply and demand balance model, the specific steps are as follows:

[0032] Based on the green electricity generation, electricity demand, grid status and meteorological conditions at the current time step, the state space S is defined. t ;

[0033] Based on the allocation ratio of green electricity, the charging and discharging behavior of energy storage, and the grid load regulation, the action space A is defined. t ;

[0034] Based on historical electricity consumption data, ARIMA is used to analyze the electricity demand of users in the future. The expression is:

[0035] D′ t+1:t+T =ARIMA(D t-L+1:t ;p,d,q);

[0036] Among them, D′ t+1:t+T It is the user electricity demand from time step t+1 to time step t+T, D t-L+1:t is the electricity consumption of the user from time step t-L+1 to time step t, p is the order of the ARIMA regression term, d is the difference order of ARIMA, q is the order of the ARIMA moving average term, and L represents the length of the historical time window;

[0037] Based on the historical supply and demand balance, the power generation in the future and the user's electricity demand, the reward function R is defined t ;

[0038] Based on the state space S t , action space A t and the reward function R t , a green electricity supply and demand balance model is constructed through a deep Q network combined with an experience replay mechanism.

[0039] As a preferred solution of the data intelligent processing method for green electricity of the present invention, based on the power generation and user electricity demand in the future period, the optimal value of the supply and demand balance is obtained through the green electricity supply and demand balance model, which is expressed as follows:

[0040]

[0041] Among them, Q(S t ,A t ; θ) represents the parameter θ of the deep Q network in state S t Next, perform action A t The expected value of the future cumulative reward is obtained, γ represents the discount factor, S t+1 represents the state space at time t+1, A t+1 represents the action space at time t+1, θ represents the learning parameters in the deep Q network, R t is the reward function;

[0042] By the expected value of future cumulative rewards Q(S t ,A t ;θ), select the optimal action A′ t ;

[0043] Based on the optimal action A′ t , obtain the optimal value of supply and demand balance;

[0044] Based on the optimal value of supply and demand balance, the optimal dispatching strategy is generated by dynamically controlling energy storage charging, grid dispatching and load regulation.

[0045] In the second aspect, the present invention provides a data intelligent processing system for green electricity, including a feature fusion module, a power generation prediction module and a strategy generation module; the feature fusion module is used to collect multi-source data, pre-process the multi-source data, extract multi-source features based on the multi-source data, and generate a comprehensive feature vector through a graph attention fusion method; the power generation prediction module is used to construct a green electricity generation prediction model, and use historical multi-source data for training, and predict the power generation in the future through the trained green electricity generation prediction model; the strategy generation module is used to define the state space, action space and reward function using a reinforcement learning algorithm, and construct a green electricity supply and demand balance model, based on the power generation and user electricity demand in the future, obtain the optimal value of the supply and demand balance, and generate the optimal scheduling strategy based on the optimal value of the supply and demand balance.

[0046] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, any step of the method for intelligent data processing for green electricity as described in the first aspect of the present invention is implemented.

[0047] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for intelligent data processing for green electricity as described in the first aspect of the present invention is implemented.

[0048] The beneficial effects of the present invention are as follows: by combining real-time meteorological data, equipment operation data, and historical electricity consumption data, and utilizing advanced feature extraction technology and graph attention fusion mechanisms to generate comprehensive feature vectors, the ability to capture the complex correlations of multi-source data is greatly improved. Subsequently, a temporal convolutional network (TCN) is used to construct and train a green electricity generation prediction model, effectively improving the accuracy and robustness of future power generation forecasts. These two steps not only optimize green electricity generation forecasts, but also enhance the intelligent management of grid supply and demand balance through dynamic adjustment strategies, achieving more efficient and stable green energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 Flowchart of the method for intelligent data processing of green electricity in Example 1.

[0051] Figure 2 Schematic diagram of the data intelligent processing system for green electricity in Example 1. DETAILED DESCRIPTION

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0055] Example 1, with reference to Figure 1and Figure 2 , which is the first embodiment of the present invention, provides a method for intelligent data processing for green electricity, comprising the following steps:

[0056] S1. Collect and preprocess multi-source data, extract multi-source features based on the multi-source data, and generate a comprehensive feature vector through graph attention fusion method.

[0057] Multi-source data includes real-time meteorological data, equipment operation data and historical electricity consumption data.

[0058] Furthermore, real-time meteorological data, including parameters such as temperature, humidity, and wind speed, is regularly collected and recorded by a network of geographically distributed sensors. Equipment operation data, including performance indicators of power generation equipment such as power output, efficiency, and fault information, is acquired and transmitted in real time by monitoring devices integrated into the equipment. Historical electricity consumption data, compiled from data regularly uploaded by smart meters, includes user consumption over different time periods. This detailed data provides a solid foundation for subsequent processing and analysis.

[0059] Preprocessing includes data cleaning, normalization, missing value processing, data alignment and denoising.

[0060] Furthermore, during the data cleaning process, a comprehensive check is conducted for erroneous values ​​and outliers in multi-source data. For each record of real-time meteorological data, equipment operation data, and historical electricity consumption data, reasonable threshold ranges are set to identify and correct or delete data points that do not conform to actual physical meaning.

[0061] Normalization aims to bring data from different sources onto the same scale for easier processing and comparison. Specifically, linear transformations are applied to each metric in real-time meteorological data, equipment operation data, and historical electricity usage data. This maps the value of each feature to a predefined interval, such as [0, 1] or [-1, 1], eliminating the effects of dimensional differences.

[0062] In the event of missing values, appropriate strategies are used to fill in the missing values ​​using real-time meteorological data, equipment operation data, and historical electricity consumption data. Depending on the specific characteristics of the missing values, methods such as mean filling, median filling, or time series forecasting are used for estimation and supplementation.

[0063] To effectively integrate multi-source data, the data alignment phase requires synchronizing real-time meteorological data, equipment operation data, and historical electricity consumption data from different timestamps to a unified time base. This requires precise timestamp matching to ensure that all related data can be compared and analyzed at the same point in time, thereby improving the accuracy of feature extraction.

[0064] Based on real-time meteorological data, meteorological features are extracted through wavelet transform and principal component analysis.

[0065] Furthermore, wavelet transforms are first applied to decompose raw meteorological data into components of different scales, thereby capturing detailed information and trends in weather changes. For example, temperature data can be decomposed into high-frequency components (such as short-term temperature fluctuations) and low-frequency components (such as diurnal or seasonal variations). Principal component analysis (PCA) is then used to reduce the dimensionality of these decomposed components, thereby extracting the most representative meteorological features.

[0066] Based on the equipment operation data, equipment operation features are extracted through time series analysis and convolutional neural networks.

[0067] Furthermore, time series analysis methods are first used to identify temporal patterns in parameters such as equipment output power. For example, by analyzing hourly output power records of wind turbines over the past month, periodicity in their operating conditions and potential fault warning signals can be detected. Next, a convolutional neural network (CNN) is used to perform deep feature learning on this time series data.

[0068] Based on historical electricity consumption data, electricity consumption characteristics are extracted through Fourier transform.

[0069] Furthermore, a Fourier transform is first used to convert a user's daily electricity consumption data from the time domain to the frequency domain. For example, considering a household's daily electricity consumption over a year, the Fourier transform can clearly show which frequency components dominate, reflecting the user's electricity usage habits (e.g., higher electricity consumption during daily peak hours).

[0070] Multi-source features are input into the support vector machine, and the meteorological features, equipment operation features and electricity consumption features are mapped to a high-dimensional space through the RBF kernel to identify the correlation between the features.

[0071] A two-level graph attention mechanism consisting of local and global levels is used to construct multi-source features into graph-structured data, where each multi-source feature is regarded as a node and the correlation between multi-source features is regarded as an edge.

[0072] It should be noted that each multi-source feature is first treated as an independent node in the graph-structured data. For example, the temperature feature in real-time meteorological data, the efficiency feature in equipment operation data, and the consumption feature in historical electricity consumption data each become a node. Next, edges are defined by analyzing the correlation between these features, such as how temperature changes affect equipment operating efficiency, or how equipment operating status changes users' electricity consumption patterns. Further processing is performed using a two-level graph attention mechanism consisting of local and global levels. At the local level, a single-layer graph attention network (GAT) is used to learn the relationship between each node and its direct neighbors, such as analyzing the impact of equipment performance on electricity consumption under specific weather conditions. At the global level, a cross-layer attention mechanism is used to explore the complex connections between nodes and indirect neighbors, such as evaluating the potential impact of long-term meteorological trends on the overall power grid load.

[0073] Based on graph structure data, at the local level, a single-layer GAT is used to learn the relationship between nodes and directly adjacent nodes, and at the global level, a cross-layer attention mechanism is used to learn the relationship between nodes and indirectly adjacent nodes.

[0074] It should be noted that a single-layer graph attention network (GAT) is first applied at the local level. This network learns the relationships between each node by calculating the attention coefficient between each node and its directly adjacent nodes. For example, in a network composed of multiple sensors, a sensor node will adjust its perception weight based on the data of its directly adjacent neighbor nodes to better reflect the characteristics of the local environment. Next, at the global level, a cross-layer attention mechanism is adopted. This process allows nodes to explore and learn the relationships with indirectly adjacent nodes. For example, in the sensor network mentioned above, even if two sensor nodes are not directly adjacent, they can still influence each other through a series of intermediate nodes. The cross-layer attention mechanism can identify and quantify these long-range dependencies, thereby enhancing the understanding and expression of the dynamics of the entire network.

[0075] It should also be noted that the cross-layer attention mechanism is a technology used in graph neural networks that aims to capture the complex relationships between nodes and their indirect neighbors. Specifically, in graph-structured data, each node is not only associated with its directly adjacent neighbor nodes, but may also affect more distant nodes through a series of intermediate nodes. The cross-layer attention mechanism quantifies this long-distance dependency by transferring information between different layers and calculating the indirect attention coefficient between each node. This process first establishes a connection between a node and its directly adjacent neighbor nodes at the local level, and then expands outward layer by layer, using the representation learned in the previous layer to update the feature representation of the node in the current layer, thereby achieving learning of the deep-level relationships of the entire graph structure. In this way, even if two nodes are not directly adjacent, the potential interaction between them can be captured through multi-step propagation, and ultimately a more comprehensive and accurate node feature representation is obtained.

[0076] Based on the relationship between the node and its direct and indirect adjacent nodes, the attention weights at different levels are fused through the multi-head attention mechanism to generate a comprehensive feature vector F, which is expressed as:

[0077]

[0078] Among them, K is the number of heads of the multi-head attention mechanism, Q k is the query vector of the kth attention head, K k is the key vector of the kth attention head, V k is the value matrix of the kth attention head, d k is the dimension of multi-source features, and T is the transpose operator.

[0079] It should be noted that the multi-head attention mechanism is an enhanced processing technology used to generate a comprehensive feature vector F based on the relationship between a node and its direct and indirect adjacent nodes. In this mechanism, multiple attention heads (K) are first defined, and each attention head independently learns the feature representation in a different subspace. Specifically, for each attention head k, by calculating the query vector Q k With the key vector K k The product of the transpose and divided by the square root of d k (the dimension of multi-source features) to measure the correlation between nodes, and then apply the softmax function to normalize these scores to obtain the attention weights between nodes. k The weighted sum is then taken based on these weights to capture the information in a specific subspace. Finally, the results of all attention heads are concatenated or averaged to form a comprehensive feature vector F.

[0080] S2. Build a green electricity generation prediction model and use historical multi-source data for training. Use the trained green electricity generation prediction model to predict power generation in the future.

[0081] Based on the statistical characteristics and distribution of historical green electricity generation, the number of TCN layers, number of neurons and learning rate are initialized.

[0082] Constructs the input layer based on the dimensions of the input data.

[0083] It should be noted that the input layer is first designed based on the dimensions of the input data. For example, if the input data includes real-time weather data, equipment operation data, and historical electricity usage data, each data type has a specific number of features. Assuming that the total number of features after combining these data is N, the input layer is designed to accept vectors of dimension N.

[0084] The hidden layers are constructed based on the initialized number of TCN layers, number of neurons, and learning rate to capture the long-term dependencies in the feature vectors in time series format.

[0085] It should be noted that when constructing a hidden layer, a hidden layer is created based on pre-set parameters such as the number of temporal convolutional network (TCN) layers, the number of neurons in each layer, and the learning rate. For example, to effectively capture the long-term dependencies in the time series feature vectors involved in green power generation forecasting, several TCN layers can be set, each containing a certain number of neurons, and an appropriate learning rate can be used to adjust the speed of parameter updates.

[0086] Use the activation function to build a fully connected layer and output the green electricity generation with consistent dimensions.

[0087] It should be noted that to convert the features learned by the hidden layer into specific green electricity generation prediction values, a fully connected layer is added to the end of the green electricity generation prediction model and an activation function is applied to this layer. For example, ReLU is chosen as the activation function, which can suppress negative values ​​while retaining positive values. This helps introduce nonlinear factors and makes the green electricity generation prediction model more expressive. The design of the fully connected layer is based on the dimensionality of the input layer, ensuring that the final output green electricity generation value matches the input dimensionality.

[0088] A green electricity generation prediction model is constructed based on the input layer, hidden layer and fully connected layer.

[0089] The green electricity generation prediction model is trained using historical multi-source data, and the mean square error is used as the loss function. Backpropagation is used to update the parameters of the green electricity generation prediction model, and the Adam optimizer is used to dynamically adjust the learning rate.

[0090] It should be noted that when training the green electricity generation prediction model, historical multi-source data, including real-time meteorological data, equipment operation data, and historical electricity consumption data, is first used as training samples and input into the green electricity generation prediction model. The prediction error is evaluated by calculating the mean squared error (MSE) between the green electricity generation prediction model output and the actual green electricity generation, and this is used as the loss function. Next, the backpropagation algorithm is used to update the parameters of the green electricity generation prediction model to minimize this loss function. During this process, the Adam optimizer automatically adjusts the learning rate based on the direction of gradient descent to ensure rapid convergence in the early stages of training. As the optimal solution is approached, the learning rate change is slowed to improve accuracy and stability. For example, when processing a year of historical data, the green electricity generation prediction model parameters are continuously adjusted to account for the difference between the predicted value and the actual power generation for each day, gradually approaching the actual value, thereby achieving an accurate prediction of green electricity generation.

[0091] The comprehensive feature vector F is divided into time steps to generate feature vectors in time series format and normalized.

[0092] The normalized time series format feature vector is input into the green electricity generation prediction model to predict the power generation in the future. The expression is:

[0093] Y t+1:t+T =ReLU(TCN(X t-L+1:t ;W1,b1))·W2+b2;

[0094] Among them, Y t+1:t+T represents the power generation from time step t+1 to time step t+T, TCN represents the temporal convolutional network, X t-L+1:t It represents the normalized feature vector in the time series format from time step t-L+1 to time step t, W1 represents the weight of the TCN layer, b1 represents the bias parameter of the TCN layer, W2 is the weight of the fully connected layer, b2 represents the bias parameter of the fully connected layer, T represents the length of the future time period, and t represents the current time.

[0095] It should be noted that when predicting the amount of green electricity generated, the normalized time series format feature vector X t-L+1:t , that is, the data from time step t-L+1 to time step t is input into the temporal convolutional network (TCN). The TCN layer uses its internal weight W1 and bias parameter b1 to process the input data to capture the long-term dependencies in the time series. The output after processing by the TCN layer introduces nonlinear factors through the ReLU activation function to enhance the expressiveness of the green electricity generation prediction model. Subsequently, the output is multiplied by the weight W2 of the fully connected layer and the bias parameter b2 is added to finally calculate the power generation Y from time step t+1 to time step t+T in the future time period t+1:t+TFor example, based on hourly weather, equipment operation, and electricity consumption data from the past week, hourly power generation for the next 24 hours can be predicted.

[0096] S3. Use reinforcement learning algorithms to define the state space, action space, and reward function, and build a green electricity supply and demand balance model. Based on the power generation and user electricity demand in the future, obtain the optimal value of the supply and demand balance, and generate the optimal scheduling strategy based on the optimal value of the supply and demand balance.

[0097] Based on the green electricity generation, electricity demand, grid status and meteorological conditions at the current time step, the state space S is defined. t .

[0098] It should be noted that the relevant information of the current time step is first collected, including the real-time generation of green electricity, the electricity demand of users, the overall state of the power grid, and the current meteorological conditions. For example, at a specific moment, assuming that the green electricity generation equipment is generating electricity according to wind speed and light intensity, the total electricity demand of users at that moment and the state of other power sources in the power grid (such as the output of thermal power stations) are recorded. In addition, the impact of meteorological parameters such as temperature and humidity on electricity demand and production must also be considered. By integrating these multi-dimensional information, a state space S that comprehensively reflects the current electricity supply and demand situation and its environmental impact is constructed. t .

[0099] Based on the allocation ratio of green electricity, the charging and discharging behavior of energy storage, and the grid load regulation, the action space A is defined. t .

[0100] It should be noted that it is necessary to determine a feasible set of actions based on how to adjust the allocation ratio of green electricity, manage the charging and discharging process of the energy storage system, and dynamically adjust the grid load. For example, consider increasing the allocation ratio of green energy when solar energy is abundant during the day, and relying on the energy storage system to supply power at night to maintain stable output. At the same time, according to the changes in the grid load, the charging and discharging strategies of the energy storage device are adjusted in time, such as charging during low-peak hours and discharging during peak hours to reduce the pressure on the grid. In addition, it may also involve balancing the overall grid load by adjusting the usage time and intensity of non-critical loads. In this way, the action space A t It covers not only operations directly related to electricity production and storage, but also various regulatory measures that indirectly affect the stability of the power grid.

[0101] Based on historical electricity consumption data, ARIMA is used to analyze the electricity demand of users in the future. The expression is:

[0102] D′ t+1:t+T =ARIMA(D t-L+1:t ;p,d,q);

[0103] Among them, D′ t+1:t+T It is the user electricity demand from time step t+1 to time step t+T, D t-L+1:t is the electricity consumption of the user from time step t-L+1 to time step t, p is the order of the ARIMA regression term, d is the difference order of ARIMA, q is the order of the ARIMA moving average term, and L represents the length of the historical time window.

[0104] Based on the historical supply and demand balance, the power generation in the future and the user's electricity demand, the reward function R is defined t .

[0105] It should be noted that an evaluation framework is first constructed based on historical supply and demand balance data, predicted power generation for a period of time in the future, and user electricity demand. Specifically, by analyzing the state of the supply and demand balance in the past, we can understand the effectiveness and efficiency of power distribution under different conditions. Then, combined with the forecast of future power generation and the estimate of user electricity demand, we can determine how to optimally dispatch power resources in the future time period to meet demand while avoiding waste. The reward function R t Based on these factors, the system is designed to maximize supply-demand matching while minimizing unmet demand or excess supply. For example, during the hot summer months, if sufficient green power generation is predicted, considering the peak electricity demand caused by air conditioning, actions that effectively utilize energy storage systems to smooth load peaks will be rewarded. Conversely, if generation is insufficient, strategies that reduce non-essential loads or optimize the allocation of limited power resources will be rewarded, thereby ensuring stable grid operation and improving overall energy efficiency.

[0106] Based on the state space S t , action space A t and the reward function R t , a green electricity supply and demand balance model is constructed through a deep Q network combined with an experience replay mechanism.

[0107] It should be noted that the experience replay mechanism is a technology used to enhance the learning process. It is built on past experience data and improves learning efficiency and effectiveness by storing and randomly sampling past decision-making steps. In this mechanism, each state transition, action taken, reward obtained and other information are recorded to form an experience pool. Subsequently, in the learning process, instead of relying solely on the most recent experience, samples are randomly drawn from the experience pool for training. This breaks the correlation between the data and makes learning more stable and efficient. For example, when optimizing the green electricity supply and demand balance model, the experience replay mechanism can be used to utilize the information about the state space S at different time steps in the past. t , action space A t And the reward function R tThe data can help the green electricity supply and demand balance model better understand the selection of optimal strategies under various circumstances, thereby improving the accuracy of forecasting and responding to future supply and demand conditions.

[0108] Based on the power generation and user electricity demand in the future, the optimal value of supply and demand balance is obtained through the green electricity supply and demand balance model, which is expressed as:

[0109]

[0110] Among them, Q(S t ,A t ; θ) represents the parameter θ of the deep Q network in state S t Next, perform action A t The expected value of the future cumulative reward is obtained, γ represents the discount factor, S t+1 represents the state space at time t+1, A t+1 represents the action space at time t+1, θ represents the learning parameters in the deep Q network, R t is the reward function.

[0111] It should be noted that, first, based on the power generation and user power demand in the future, the green power supply and demand balance model is used to evaluate the effects of different strategies. Specifically, for each time step t, the state space S t and action space A t , the expected value Q(S) of the future cumulative reward after performing a specific action is calculated through the deep Q network parameter θ t ,A t ;θ). During this calculation process, the reward value R at the current moment t It is added to the maximum possible Q value at the next moment, which is weighted by the discount factor γ, and is then used to find the next state S t+1 Next action space A t+1 The action that produces the maximum Q value is implemented. This allows for iterative updates of the estimated value of each action in each state until the optimal strategy that maximizes future cumulative rewards is found, ensuring the best match between electricity supply and demand. For example, in the face of an upcoming peak in electricity demand, this mechanism can help determine the most appropriate release of energy storage or increase the proportion of other power sources to achieve the optimal balance between supply and demand.

[0112] By the expected value of future cumulative rewards Q(S t ,A t ;θ), select the optimal action A′ t .

[0113] Based on the optimal action A′ t , to obtain the optimal value of supply and demand balance.

[0114] It should be noted that in determining the optimal action A′ t In the process of , we first calculate the expected value Q(S) of the future cumulative reward of each possible action. t ,A t ;θ), based on the current state space S t and action space A t By comparing these expected values, the action that produces the highest future cumulative reward is selected as the optimal action A′ t Specifically, for each state S at time step t t , evaluate all possible actions A t The corresponding Q value is obtained, and the action with the maximum Q value is selected for execution. Next, according to the selected optimal action A′ t This process adjusts the green power allocation ratio, energy storage charging and discharging behavior, and grid load regulation to achieve the optimal balance between supply and demand. This process ensures that, in any given state, actions taken are directed toward maximizing long-term rewards, ultimately achieving the goal of optimizing power resource allocation and maintaining grid stability. For example, if electricity demand is predicted to exceed green power supply in the future, the system can increase energy storage output or adjust the operating hours of non-critical loads to maintain the optimal balance between supply and demand.

[0115] Based on the optimal value of supply and demand balance, the optimal dispatching strategy is generated by dynamically controlling energy storage charging, grid dispatching and load regulation.

[0116] Furthermore, strategies for when supply exceeds demand include: Energy storage charging: storing excess green electricity in energy storage systems. Grid dispatching: delivering excess green electricity to other regions or the grid. Load regulation: increasing power supply to low-priority loads.

[0117] Strategies for when supply falls short of demand: Energy storage discharge: Releasing energy from the storage system to fill the gap. Grid dispatch: Transferring energy from other regions or the grid. Load regulation: Reducing power supply to low-priority loads.

[0118] Strategy when supply and demand are balanced: Maintain the current dispatch strategy and optimize the efficiency of energy storage systems and power grid operations.

[0119] This embodiment also provides a data intelligent processing system for green electricity, including: a feature fusion module, a power generation prediction module and a strategy generation module; the feature fusion module is used to collect multi-source data, pre-process the multi-source data, extract multi-source features based on the multi-source data, and generate a comprehensive feature vector through a graph attention fusion method; the power generation prediction module is used to construct a green electricity generation prediction model, and use historical multi-source data for training, and predict the power generation in the future through the trained green electricity generation prediction model; the strategy generation module is used to define the state space, action space and reward function using a reinforcement learning algorithm, and construct a green electricity supply and demand balance model, obtain the optimal value of the supply and demand balance based on the power generation and user electricity demand in the future, and generate the optimal scheduling strategy based on the optimal value of the supply and demand balance.

[0120] This embodiment also provides a computer device suitable for the data intelligent processing method for green electricity, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the data intelligent processing method for green electricity proposed in the above embodiment.

[0121] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0122] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the program implements the data intelligent processing method for green electricity proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0123] In summary, the present invention combines real-time meteorological data, equipment operation data, and historical electricity consumption data, and utilizes advanced feature extraction technology and graph attention fusion mechanisms to generate comprehensive feature vectors, thereby greatly improving the ability to capture the complex correlations of multi-source data. Subsequently, a temporal convolutional network (TCN) is used to construct and train a green electricity generation prediction model, effectively improving the accuracy and robustness of future power generation forecasts. These two steps not only optimize the green electricity generation forecast, but also enhance the intelligent management of the power grid supply and demand balance through dynamic adjustment strategies, achieving more efficient and stable green energy utilization.

[0124] Example 2, referring to Table 1, is the second embodiment of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the data intelligent processing method for green electricity are provided.

[0125] To validate the innovativeness and advantages of intelligent multi-source data processing in optimizing green electricity supply and demand, the experiment first collected 30 days of real-time meteorological data, equipment operation data, and historical electricity consumption data. Preprocessing included data cleaning, normalization, missing value processing, data alignment, and denoising. Outliers were identified and corrected by setting reasonable thresholds, and missing values ​​were filled using time series forecasting methods.

[0126] Subsequently, multi-source features are extracted using techniques such as wavelet transform, principal component analysis, time series analysis, convolutional neural networks, and Fourier transform. Support vector machines and RBF kernels are then used to map these features into a high-dimensional space to identify correlations between features. Next, a two-level graph attention mechanism is applied to construct graph-structured data. A single-layer GAT and a cross-layer attention mechanism are used to learn relationships between local and global levels, ultimately generating a comprehensive feature vector F.

[0127] Based on the comprehensive feature vector F, a green power generation prediction model consisting of a TCN layer and a fully connected layer was constructed. This model was trained using historical multi-source data. A deep Q-network combined with an experience replay mechanism was used to evaluate the effectiveness of different strategies and select the optimal action to achieve the optimal value for supply and demand balance. This process ensures accurate predictions of future power generation and the development of efficient scheduling strategies.

[0128] The existing technologies discussed in this experiment employ traditional green electricity supply and demand balancing strategies based on statistics and simple machine learning methods. These include basic time series analysis for power generation forecasting, simple linear regression or decision tree models for processing multi-source data, and grid dispatch and load regulation mechanisms that rely on fixed rules or thresholds.

[0129] The details are shown in Table 1:

[0130] Table 1 Comparison of experimental data on green electricity supply and demand balance optimization

[0131]

[0132] Analyzing the data in the table above, it's clear that the present invention significantly outperforms existing solutions in several key metrics. For example, in terms of power generation forecast accuracy, the present invention achieved 91.8%, while the existing technology achieved only 74.2%. This demonstrates that the present invention can more accurately predict future power generation, providing a more reliable foundation for grid scheduling. Furthermore, the present invention's scheduling strategy effectiveness reached 84.7%, compared to only 61.5% for the existing technology. This means that the present invention not only improves forecast accuracy but also enables more efficient scheduling strategies, ensuring an optimal match between power supply and demand.

[0133] In summary, the present invention has demonstrated significant advantages in terms of technical performance, economic benefits and user experience, and the novelty of the present invention is particularly evident in extreme situations.

[0134] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for intelligent data processing of green electricity, characterized by: include, Collect and preprocess multi-source data, extract multi-source features based on the multi-source data, and generate a comprehensive feature vector through the graph attention fusion method. The specific steps are as follows: A two-level graph attention mechanism consisting of local and global levels is used to construct multi-source features into graph-structured data, where each multi-source feature is regarded as a node and the correlation between multi-source features is regarded as an edge. Based on graph structure data, at the local level, a single-layer GAT is used to learn the relationship between nodes and directly adjacent nodes, and at the global level, a cross-layer attention mechanism is used to learn the relationship between nodes and indirectly adjacent nodes; Based on the relationship between the node and its direct and indirect adjacent nodes, the multi-head attention mechanism is used to fuse the attention weights of different levels to generate a comprehensive feature vector ; The multi-source data includes real-time meteorological data, equipment operation data and historical electricity consumption data; Build a green electricity generation prediction model and use historical multi-source data for training. Use the trained green electricity generation prediction model to predict power generation in the future. The specific steps are as follows: Initialize the number of TCN layers, number of neurons, and learning rate based on the statistical characteristics and distribution of historical green power generation; Construct an input layer based on the dimensions of the input data; Construct hidden layers based on the initialized number of TCN layers, number of neurons, and learning rate to capture long-term dependencies in the feature vectors in time series format; Use activation functions to construct a fully connected layer to output the green electricity generation amount with consistent dimensions; Build a green electricity generation prediction model based on the input layer, hidden layer and fully connected layer; The green electricity generation prediction model is trained using historical multi-source data, and the mean square error is used as the loss function. Backpropagation is used to update the parameters of the green electricity generation prediction model, and the Adam optimizer is used to dynamically adjust the learning rate. Divide the comprehensive feature vector F into time steps, generate feature vectors in time series format, and perform normalization processing; The normalized time series format feature vector is input into the green electricity generation prediction model to predict the power generation in the future. The expression is: ； in, Represents the time step To time step of power generation, represents a temporal convolutional network, Indicates that at time step To time step The feature vector in the normalized time series format, express The weights of the layers, express The bias parameters of the layer, is the weight of the fully connected layer, represents the bias parameter of the fully connected layer, Indicates the length of the future time period, Indicates the current time; A reinforcement learning algorithm is used to define the state space, action space and reward function, and a green electricity supply and demand balance model is constructed. Based on the power generation and user electricity demand in the future, the optimal value of the supply and demand balance is obtained, and the optimal scheduling strategy is generated based on the optimal value of the supply and demand balance.

2. The method for intelligent processing of green electricity data according to claim 1, characterized in that: The preprocessing includes data cleaning, normalization, missing value processing, data alignment and denoising.

3. The method for intelligent processing of green electricity data according to claim 2, wherein: The method extracts multi-source features based on multi-source data and generates a comprehensive feature vector through the graph attention fusion method. The specific steps are as follows: Based on real-time meteorological data, meteorological features are extracted through wavelet transform and principal component analysis; Based on the equipment operation data, the equipment operation characteristics are extracted through time series analysis and convolutional neural network; Based on historical electricity consumption data, electricity consumption characteristics are extracted through Fourier transform; Multi-source features are input into the support vector machine, and meteorological features, equipment operation features, and electricity consumption features are mapped into a high-dimensional space using the RBF kernel to identify the correlation between features. Generate comprehensive feature vector , the expression is: ； in, is the number of heads of the multi-head attention mechanism, It is The query vector of the attention head, It is The key vector of the attention head, It is The value matrix of the attention head, is the dimension of multi-source features, and T is the transpose operator.

4. The method for intelligent processing of green electricity data according to claim 3, wherein: The reinforcement learning algorithm is used to define the state space, action space and reward function, and to build a green electricity supply and demand balance model. The specific steps are as follows: Based on the green electricity generation, electricity demand, grid status and meteorological conditions at the current time step, the state space is defined. ; Define the action space based on the green electricity allocation ratio, energy storage charging and discharging behavior, and grid load regulation ; Based on historical electricity consumption data, ARIMA is used to analyze the electricity demand of users in the future. The expression is: ； in, is in the time step To time step The electricity demand of users, is in the time step To time step The electricity consumption of users, yes The order of the regression term, yes The difference order of yes The order of the moving average term, Indicates the length of the historical time window; Define the reward function based on the historical supply and demand balance, the power generation in the future and the user's electricity demand ; Based on state space , action space and the reward function , a green electricity supply and demand balance model is constructed through a deep Q network combined with an experience replay mechanism.

5. The method for intelligent processing of green electricity data according to claim 4, characterized in that: Based on the power generation and user electricity demand in the future, the optimal value of supply and demand balance is obtained through the green electricity supply and demand balance model, which is expressed as: ； in, Represents the parameters of the deep Q network In state Next action The expected value of the future cumulative rewards obtained, represents the discount factor, express The state space at time, express The action space at each moment, represents the learning parameters in the deep Q network, is the reward function; By the expected value of future accumulated rewards , choose the best action ; Based on optimal action , obtain the optimal value of supply and demand balance; Based on the optimal value of supply and demand balance, the optimal dispatching strategy is generated by dynamically controlling energy storage charging, grid dispatching and load regulation.

6. A data intelligent processing system for green electricity, based on the data intelligent processing method for green electricity according to any one of claims 1 to 5, characterized in that: Including feature fusion module, power generation prediction module and strategy generation module; The feature fusion module is used to collect and preprocess multi-source data, extract multi-source features based on the multi-source data, and generate a comprehensive feature vector through the graph attention fusion method; The power generation prediction module is used to build a green power generation prediction model and use historical multi-source data for training. The trained green power generation prediction model is used to predict power generation in the future. The strategy generation module is used to define the state space, action space, and reward function using a reinforcement learning algorithm, and to build a green electricity supply and demand balance model. Based on the power generation and user electricity demand over a period of time in the future, it obtains the optimal value of the supply and demand balance and generates the optimal scheduling strategy based on the optimal value of the supply and demand balance.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for intelligent processing of green electricity data according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for intelligent data processing for green electricity according to any one of claims 1 to 5 are implemented.

Images