Wind power prediction driven power grid energy management method, system and device

Abstract

This invention discloses a wind power forecast-driven grid energy management method, system, and equipment, belonging to the field of grid optimization and dispatching technology. The method includes: acquiring power forecast distribution sequences and confidence interval data of wind farms over multiple forecast periods; constructing an uncertain energy evolution tensor containing temporal gradients, spatial diffusion, and probabilistic discrete components; mapping this tensor to the grid node topology to generate a risk seepage topology map; identifying key channels and nodes and constructing a risk seepage channel set; subsequently calculating a partitioned elasticity threshold to generate an elastic risk envelope region; and finally, combining stored grid energy to perform hierarchical energy regulation and allocation, establishing a grid energy management strategy. This invention solves the technical problem that the intermittency and uncertainty of wind power easily lead to energy dispatching risks in distribution circuit systems, achieving accurate risk identification and dynamic control within the grid, and combining grid energy storage to complete differentiated energy regulation and allocation.

Description

Technical Field

[0001] This invention relates to the field of power grid optimization and dispatching technology, and in particular to a wind power forecast-driven power grid energy management method, system, and equipment. Background Technology

[0002] The power distribution circuit system is a core component of power supply, and its energy management effectiveness directly impacts grid stability and energy utilization efficiency. In the context of wind power grid integration, precise energy management is crucial for the safe and efficient operation of the power distribution circuit system. Existing technologies largely employ conventional dispatching methods, combining baseline power forecasting with energy allocation and control of the power distribution circuit system. While these methods have proven effective under stable operating conditions, their limitations are becoming increasingly apparent as wind power penetration increases. Due to the intermittent and uncertain nature of wind power generation, traditional methods cannot quantify the propagation patterns of wind power uncertainty, resulting in a lack of dynamic adaptability in energy management strategies, incomplete data support, and an inability to meet the refined and adaptive energy management and risk control requirements of the power distribution circuit system. Summary of the Invention

[0003] This application provides a wind power forecast-driven grid energy management method, system, and equipment, which solves the technical problem that the intermittency and uncertainty of wind power can easily lead to energy dispatch risks in the distribution circuit system.

[0004] The first aspect of this application provides a wind power prediction-driven grid energy management method, the method comprising: acquiring power prediction distribution sequences and corresponding confidence interval data of wind farms in multiple prediction periods; constructing an uncertain energy evolution tensor containing temporal gradient components, spatial diffusion components, and probabilistic discrete components based on the acquisition results; mapping the uncertain energy evolution tensor to the grid node topology, calculating the risk coupling strength coefficient and uncertain energy seepage direction parameters of each grid node, and generating a risk seepage topology map characterizing the propagation trend of uncertainty in the grid; identifying high-density channels and key coupling nodes of uncertain energy based on the risk seepage topology map, and constructing a risk seepage channel set; calculating a partition elastic threshold on the risk seepage channel set based on the temporal change rate and spatial diffusion coefficient of the uncertain energy evolution tensor, and generating an elastic risk envelope region with adaptively adjustable boundaries; and performing hierarchical energy regulation and allocation based on stored grid energy and the elastic risk envelope region to establish a grid energy management strategy.

[0005] A second aspect of this application provides a wind power prediction-driven grid energy management system, the system comprising: an energy evolution tensor construction module, used to acquire power prediction distribution sequences and corresponding confidence interval data of wind farms in multiple prediction periods, and construct an uncertain energy evolution tensor containing time gradient components, spatial diffusion components, and probabilistic discrete components based on the acquisition results; a risk permeation topology map acquisition module, used to map the uncertain energy evolution tensor to the grid node topology, calculate the risk coupling strength coefficient and uncertain energy permeation direction parameters of each grid node, and generate a risk permeation topology map characterizing the propagation trend of uncertainty in the grid; a risk permeation channel set construction module, used to identify high-density channels and key coupling nodes of uncertain energy based on the risk permeation topology map, and construct a risk permeation channel set; a flexible risk envelope region acquisition module, used to calculate the partitioning elastic threshold on the risk permeation channel set based on the time change rate and spatial diffusion coefficient of the uncertain energy evolution tensor, and generate a flexible risk envelope region with adaptively adjustable boundaries; and a grid energy management strategy construction module, used to perform hierarchical energy regulation and allocation based on stored grid energy and the flexible risk envelope region, and establish a grid energy management strategy.

[0006] A third aspect of this application provides an electronic device comprising: a memory for storing executable instructions; and a processor for implementing a wind power forecast-driven grid energy management method when executing the executable instructions stored in the memory.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] This application collects wind farm power prediction distribution and confidence interval data over multiple prediction periods. Through tensor construction, topology mapping, and risk identification, it obtains grid risk coupling and seepage-related data, calculates the zonal elasticity threshold and target energy allocation for each risk zone, and adjusts the data based on the dynamic delineation of the elastic risk envelope and the grid's stored energy. This enables precise energy management of the distribution circuit system in wind power grid-connected scenarios, making wind power prediction-driven grid energy management decisions more scientific and efficient. It achieves the technical effect of accurate risk identification and dynamic control within the grid, combined with grid energy storage to complete differentiated energy regulation and allocation. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating the wind power prediction-driven grid energy management method provided in the embodiments of this application.

[0011] Figure 2 This is a schematic diagram of the structure of the wind power prediction-driven grid energy management system provided in the embodiments of this application.

[0012] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0013] Figure labeling: Energy evolution tensor construction module 1, risk seepage topology map acquisition module 2, risk seepage channel set construction module 3, elastic risk envelope acquisition module 4, power grid energy management strategy construction module 5, input device 301, processor 302, memory 303, output device 304. Detailed Implementation

[0014] This application provides a wind power forecast-driven grid energy management method, system, and equipment, which solves the technical problem that the intermittency and uncertainty of wind power can easily lead to energy dispatch risks in the distribution circuit system.

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0016] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0017] Example 1, as Figure 1 As shown, a wind power forecast-driven grid energy management method is described, wherein the method includes:

[0018] Obtain the power prediction distribution sequence and corresponding confidence interval data of wind farms in multiple prediction periods, and construct an uncertain energy evolution tensor containing temporal gradient components, spatial diffusion components and probabilistic discrete components based on the obtained results.

[0019] Specifically, firstly, historical and real-time monitoring data of the target wind farm are collected. This data covers meteorological parameters, unit operating status parameters, and historical real-time power data for each wind turbine within the wind farm. The data sampling interval is set to 5 to 15 minutes. The collected raw data is then cleaned to remove missing and outlier values. Linear interpolation combined with a moving average method is used to complete the missing data. Wavelet transform is then used to perform multi-scale decomposition of the completed time series data. The db4 wavelet is selected as the wavelet basis function, and the decomposition level is set to 4 levels to separate high-frequency fluctuation components and low-frequency trend components. Feature signals of different frequency components are extracted. Then, the ICEEMDAN algorithm is used to perform secondary mode decomposition on each component after wavelet decomposition. The algorithm noise standard deviation is set to 0.2, and the ensemble iteration is set to 100 times to extract the intrinsic mode functions corresponding to each component. Min-max normalization is performed on all intrinsic mode functions to map the data to the interval between 0 and 1. The preprocessed data is then divided into training and test sets using the time sliding window method. The step size of the time sliding window is consistent with the prediction period length set later. The prediction period is set to 15 minutes per period, and the prediction duration covers 0 to 4 hours.

[0020] Next, the wind power time-series feature data that has been preprocessed and normalized is used as the model input. The dimension of each input sample is [time sliding window step size, feature dimension]. The time sliding window step size is consistent with the duration of the prediction period. The feature dimension covers meteorological parameters after wavelet transform and ICEEMDAN decomposition, unit operating status parameters, and multi-scale intrinsic mode function features corresponding to historical power data. A wind power prediction model is constructed based on a Long Short-Term Memory (LSTM) network. The main network structure is a two-layer bidirectional LSTM structure, with each layer containing forward and backward propagation links and 128 hidden nodes. Each LSTM unit consists of four core modules: a forget gate, an input gate, a cell state update unit, and an output gate. The sigmoid function is used for the gating activation of the forget gate, input gate, and output gate, while the tanh function is used for the activation of the cell state candidate values. A dropout layer is set at the output of each bidirectional LSTM structure, with a fixed dropout rate of 0.2 to suppress model overfitting. The output of the two-layer bidirectional LSTM structure is concatenated and then fed into a fully connected layer with 64 neurons and the ReLU function as the activation function. Finally, the wind power prediction value for the corresponding prediction period is output through the output layer mapping. The particle swarm optimization algorithm is used to globally optimize the core hyperparameters of the prediction model, such as the learning rate, number of iterations, and number of neurons in the fully connected layer. The algorithm population size is set to 30, the inertia weight is set to 0.7, the individual learning factor and the global learning factor are set to 1.5 and 2.0, respectively, and the maximum number of iterations is set to 100. The globally optimal parameter combination is locked within the learning rate range of 0.0001 to 0.01, the number of iterations range of 50 to 300, and the number of neurons in the fully connected layer range of 32 to 128, thus completing the final construction of the prediction model.

[0021] The preprocessed training set data is input into the constructed prediction model. The actual wind power value corresponding to the training set samples is used as the fitting label. The Adam optimizer combined with the cosine annealing learning rate strategy is used to iteratively train the model. The mean squared error function is selected as the loss function, the batch size is set to 256, the initial learning rate is the global optimum locked by the particle swarm optimization algorithm, the period of the cosine annealing strategy is set to 20 epochs, the lower limit of the learning rate is set to 0.00001, the maximum training period is set to 300 epochs, and an early stopping mechanism is set with an early stopping threshold of 50 epochs. When the validation set loss value does not show an effective decrease for 50 consecutive epochs, the model training is terminated early to avoid overfitting. During training, the prediction accuracy of the trained model is simultaneously verified using test set data. Four indicators—root mean square error (RMSE), mean absolute error (MAE), symmetric mean absolute percentage error (SAb) and coefficient of determination (CDE)—are used to comprehensively evaluate model performance. For example, the preset accuracy requirements are: RMSE ≤ 3MW, MAE ≤ 2.8MW, SAb ≤ 5%, and CDE ≥ 0.92. The training process is repeated iteratively until the model output meets all preset accuracy requirements. Real-time monitoring data of the target wind farm is processed using the same preprocessing procedure as the training set and then input into the trained prediction model. The model outputs wind power prediction values ​​for multiple consecutive prediction periods, arranged chronologically to form a power prediction distribution sequence. Simultaneously, the distribution pattern of the model's historical prediction errors is statistically analyzed to verify that it conforms to a normal distribution. Based on the mean and standard deviation of this normal distribution, the upper and lower limits of the confidence intervals at 90%, 95%, and 99% confidence levels are calculated for the power prediction values ​​in each prediction period. Finally, the power prediction distribution sequence and corresponding confidence interval data are obtained.

[0022] For the obtained power prediction distribution sequence, the first-order backward difference method is used to calculate the power change rate of wind power in each prediction period. The sliding calculation window is set to the duration of a single prediction period. The trend features of wind power change over time are extracted. The calculated power change rate is normalized to form the time gradient component of the uncertain energy evolution tensor. The dimension of the time gradient component is consistent with the number of prediction periods.

[0023] For the distribution circuit system connected to the target wind farm, the electrical connection relationships and branch impedance matrices of each wind farm access node and surrounding distribution nodes in the grid node topology are obtained. The node impedance matrix is ​​obtained by inverting the branch impedance matrix. The equivalent electrical distance between the wind farm access node and each distribution node is calculated using the ratio of mutual impedance to self-impedance in the node impedance matrix. Combining the spatial distribution characteristics of the wind power prediction distribution sequence, the spatial transmission coefficient of wind power fluctuations between different distribution nodes is calculated based on the reciprocal of the equivalent electrical distance. The spatial diffusion characteristics of power fluctuations are extracted. The calculated spatial transmission coefficients are normalized using min-max normalization to map the values ​​to the interval between 0 and 1, forming the spatial diffusion component of the uncertain energy evolution tensor. The dimension of the spatial diffusion component is consistent with the number of grid nodes.

[0024] For the acquired confidence interval data, the probability distribution of power prediction is discretized hierarchically according to three confidence levels: 90%, 95%, and 99%, resulting in discrete probability nodes for three confidence levels. The probability mass sum within each confidence level set is calculated; this sum is the integral value of the probability density function within the corresponding confidence interval. The probability deviation parameter for each confidence level set is calculated; this deviation is the ratio of the mean square error between the predicted and measured values ​​within the corresponding confidence interval to the interval's half-width. A normalized weight function is constructed based on the probability mass sum and probability deviation parameter. The probability weight value corresponding to each discrete probability node is calculated, and the probability distribution features corresponding to each discrete probability node are extracted to form the discrete probability components of the uncertain energy evolution tensor. The dimension of the discrete probability components is consistent with the number of confidence levels.

[0025] Finally, the extracted temporal gradient component, spatial diffusion component, and probabilistic discrete component are subjected to dimension matching and tensor integration operations. A three-dimensional uncertain energy evolution tensor is constructed with the time dimension as the first dimension, the spatial dimension as the second dimension, and the probabilistic dimension as the third dimension. Each element within the tensor corresponds to the quantized value of wind power uncertain energy at a specific prediction period, a specific grid node, and a specific confidence level. This quantized value is obtained by weighted summation of the temporal gradient component, spatial diffusion component, and probabilistic discrete component values ​​in the corresponding dimension, with each weight coefficient set to 1 / 3. This completes the construction of the uncertain energy evolution tensor.

[0026] The uncertainty energy evolution tensor is mapped to the grid node topology, and the risk coupling strength coefficient and uncertainty energy seepage direction parameters of each grid node are calculated to generate a risk seepage topology map that characterizes the propagation trend of uncertainty in the grid.

[0027] In this embodiment, the risk coupling strength coefficient is a quantitative value representing the degree of risk impact of wind power uncertainty transmitted between adjacent nodes and branches in the distribution circuit system. It is used to characterize the strength of risk coupling correlation and the magnitude of risk transmission gain between grid nodes caused by wind power fluctuations. The uncertainty energy seepage direction parameter is a quantitative representation of the main propagation path and transmission direction of grid risks caused by wind power uncertainty in the distribution circuit system topology network. It is used to clarify the diffusion direction and transmission trend of uncertainty energy brought about by wind power fluctuations in the grid.

[0028] Optionally, the temporal gradient component of the uncertain energy evolution tensor is projected onto the power injection change of the grid nodes, the spatial diffusion component, and the grid branch impedance matrix and power flow sensitivity matrix for coupling calculation to construct the node uncertain energy influence matrix. After being weighted and corrected by multi-level probability weight factors obtained by hierarchical discretization of probabilistic discrete components, the uncertainty energy transfer gain between nodes is calculated to form the risk coupling strength coefficient matrix. Then, the uncertain energy seepage direction parameters of multiple scenarios are extracted through conditional gradient divergence calculation. Finally, with the risk coupling strength coefficient as the edge weight and the uncertain energy seepage direction of multiple scenarios as the directed constraint, a multi-probability hierarchical weighted directed risk seepage topology map representing the propagation trend of uncertainty in the power grid is constructed. This step will be explained in detail in the following content.

[0029] Based on the risk seepage topology map, identify high-density channels with uncertain energy and key coupling nodes, and construct a risk seepage channel set.

[0030] In one embodiment of this application, in the risk seepage topology graph, the node seepage intensity index is first calculated by combining the risk coupling strength coefficient of the edge weights with the seepage direction parameters. Then, the path cumulative gain operation is performed on the topology graph, and the directed path with the cumulative value exceeding the preset seepage threshold is set as an uncertain energy high-density channel. Next, the node flow concentration index is calculated based on the node seepage intensity index and the path cumulative gain result to identify key coupling nodes. Finally, the corresponding high-density channels and key coupling nodes are identified in different probability level scenarios. The risk seepage channel set is constructed through intersection enhancement and union expansion analysis. This step will be explained in detail in the following content.

[0031] Based on the time change rate of the uncertain energy evolution tensor and the spatial diffusion coefficient, the partition elastic threshold is calculated on the set of risk seepage channels to generate an elastic risk envelope region with adaptively adjustable boundaries.

[0032] Specifically, firstly, the temporal rate of change and spatial diffusion coefficient of the uncertain energy evolution tensor are extracted for each channel in the risk seepage channel set. The channel risk evolution index is constructed by combining the cumulative coupling strength of the path. Then, a dynamic threshold function is constructed based on the weighted combination of the temporal rate of change and the spatial diffusion coefficient. Subsequently, the comprehensive risk value of each node in the risk seepage topology is calculated and compared with the output value of the dynamic threshold function to configure the risk envelope boundary. Finally, during the continuous prediction period, the risk envelope boundary is expanded or contracted according to whether the increment of the channel risk evolution index exceeds the preset change ratio, forming an elastic risk envelope region with adaptively adjustable boundaries. This step will be explained in detail in the following content.

[0033] A grid energy management strategy is established by performing hierarchical energy regulation and allocation based on the stored grid energy and the elastic risk envelope.

[0034] Specifically, firstly, within the elastic risk envelope, the grid nodes are divided into high, medium, and low risk zones, and the risk weight coefficient is determined by combining the number of nodes in each risk zone with the risk evolution index. Then, based on the stored grid energy, an energy storage available energy matrix is ​​constructed and coupled with the risk weight coefficient to calculate the target energy allocation for each risk zone. Finally, a hierarchical energy regulation and allocation with high-medium-low risk constraints is performed according to the target energy allocation to establish a grid energy management strategy. This step will also be explained in detail later.

[0035] Furthermore, the method provided in this application embodiment includes:

[0036] The time gradient component in the uncertain energy evolution tensor is projected onto the power injection change of the grid nodes. The spatial diffusion component is coupled with the grid branch impedance matrix and power flow sensitivity matrix to construct the node uncertain energy influence matrix. The confidence interval hierarchical discretization processing is performed on the probability discrete component to obtain multi-level probability weighting factors. These multi-level probability weighting factors are introduced into the node uncertain energy influence matrix to weight and correct the uncertainty energy influence amplitude of each node, generating a probability-modulated node uncertain energy influence matrix. Based on the probability-modulated node uncertain energy influence matrix and the electrical distance coefficient between grid nodes, the uncertainty energy transfer gain between nodes is calculated to form a risk coupling strength coefficient matrix. Conditional gradient divergence is performed on the risk coupling strength coefficient matrix at different probability levels to extract the main diffusion direction vector of uncertain energy under different probability scenarios, thereby determining the uncertain energy seepage direction parameters in multiple scenarios. Using the risk coupling strength coefficient as the edge weight and the uncertain energy seepage direction in multiple scenarios as the directed constraint, a multi-probability level weighted directed risk seepage topology is constructed.

[0037] Specifically, the basic topology parameters of the target power distribution circuit system are first obtained. Then, the grid branch impedance matrix is ​​calculated using the nodal admittance matrix solution method for the power system. Next, the power flow sensitivity partial derivatives are calculated based on the Newton-Raphson method to obtain the partial derivatives of active power with respect to voltage phase angle and reactive power with respect to voltage amplitude at each node, forming the power flow sensitivity matrix. For the time gradient component in the uncertain energy evolution tensor, a linear projection mapping method is used. The values ​​of the time gradient components corresponding to each prediction period are assigned to the node locations of the wind farm connected to the grid, corresponding to the power injection change of that node within the corresponding period, thus completing the mapping of time-dimensional uncertainty to grid nodes.

[0038] Next, the spatial diffusion component in the uncertain energy evolution tensor is coupled with the grid branch impedance matrix and the power flow sensitivity matrix through matrix multiplication. During the operation, the spatial diffusion component is used as the base coefficient and multiplied by the node mutual impedance values ​​in the branch impedance matrix and the corresponding node sensitivity coefficients in the power flow sensitivity matrix to obtain the spatial coupling influence coefficients between nodes. The node power injection change and the spatial coupling influence coefficients are then matrix-superimposed to construct the initial node uncertain energy influence matrix. The row dimension of the matrix corresponds to the grid node number, and the column dimension corresponds to the number of prediction periods. Each element in the matrix represents the initial magnitude of the uncertain energy influence of the corresponding node within the corresponding prediction period.

[0039] Then, based on the confidence level interval of each discrete probability node in the discrete probability component, it is divided into at least three confidence level sets. The probability quality sum and probability deviation parameter of each set are then calculated. Based on the parameter, a normalized weight function is constructed and the corresponding multi-level probability weight factor is calculated. This step will be explained in detail in the following content.

[0040] Next, the multi-level probability weighting factors corresponding to the three confidence levels are matched with the elements of the corresponding confidence levels in the initial node uncertainty energy influence matrix. During the matching process, the weighting factors are mapped one-to-one with the matrix elements according to the prediction time period dimension. Using a matrix element weighting operation method, the probability weighting factor corresponding to each confidence level is multiplied by the element value of the corresponding node and prediction time period in the node uncertainty energy influence matrix. This performs element-wise weighted correction on the uncertainty energy influence amplitude of each node. During the correction process, the weighting factors corresponding to higher confidence levels are assigned higher amplitude correction weights, and the weighting factors corresponding to lower confidence levels are assigned lower amplitude correction weights. The matrix elements after weighted correction for the three confidence levels are superimposed and integrated to generate a probability-modulated node uncertainty energy influence matrix. Each element in the matrix incorporates uncertainty characteristics from three dimensions: temporal gradient, spatial diffusion, and probability discretization, comprehensively characterizing the comprehensive impact of wind power uncertainty on each node of the power grid.

[0041] Then, the probability-modulated node uncertainty energy influence matrix is ​​multiplied by the inverse of the equivalent electrical distance between grid nodes, and the node pair transfer gain function is constructed by combining the power flow sensitivity partial derivative coefficient, thereby generating the risk coupling strength coefficient matrix. This step will be explained in detail later.

[0042] Subsequently, by performing eigenvalue decomposition on the risk coupling strength coefficient matrix at each probability level, the eigenvector corresponding to the largest eigenvalue is extracted as the main diffusion direction vector of the uncertainty energy in the corresponding probability scenario, and the seepage direction parameter is determined based on the sign of the eigenvector. This step will also be explained in detail later.

[0043] Finally, a directed graph construction method from graph theory is adopted, using each grid node in the power distribution circuit system as the vertices of the topology graph. The number of vertices is exactly the same as the total number of grid nodes, and each vertex corresponds to a unique grid node number. The calculated risk coupling strength coefficient is used as the edge weight connecting the corresponding vertices in the topology graph. The edge weight value is positively correlated with the risk coupling strength coefficient and is used to quantify the strength of risk transmission between two nodes. The multi-scenario uncertain energy seepage direction parameter is used as the directed constraint for the corresponding edges in the topology graph. Based on the energy inflow and outflow directions determined by the seepage direction parameter, the edge direction is set to point from the energy outflow node to the energy inflow node, forming weighted directed edges. For the three probability levels of 99%, 95%, and 90%, weighted directed subgraphs are constructed for each level. The subgraphs of the three probability levels are then hierarchically integrated so that each directed edge is synchronously associated with the edge weight values ​​under the three probability levels. Finally, a multi-probability level weighted directed risk permeation topology graph is constructed. The vertices, directed edges, and edge weights in this topology graph correspond to grid nodes, risk transmission paths, and risk coupling strength, respectively, which fully characterizes the propagation trend of wind power uncertainty in the grid.

[0044] Through a series of steps including electrical distance coupling computation, multi-probability hierarchical matrix eigenvalue decomposition, and directed graph theory modeling, the propagation trend of wind power uncertainty in the power grid was accurately quantified and visualized, providing core topological data support for subsequent risk channel identification and power grid energy management strategy formulation.

[0045] Furthermore, the method provided in this application embodiment includes:

[0046] Based on the confidence level intervals corresponding to each discrete probability node in the discrete probability component, the discrete probability nodes are divided into at least three confidence level sets; the sum of probability quality and probability deviation parameters within each confidence level set are statistically analyzed; a normalized weight function is constructed based on the sum of probability quality and probability deviation parameters of each confidence level set, and the corresponding multi-level probability weight factors are calculated.

[0047] Optionally, the discrete probability components in the uncertainty energy evolution tensor are first extracted to obtain the confidence level interval values ​​corresponding to each discrete probability node. According to the order of confidence level from high to low, the discrete probability nodes are divided into three confidence level sets, corresponding to the three confidence level intervals of 99%, 95%, and 90%, respectively. Each confidence level set contains all discrete probability node data under the corresponding confidence level.

[0048] For each confidence level set, the probability density function integral method is used to calculate the definite integral value of the probability density function within the confidence interval, obtaining the sum of probability quality for the corresponding confidence level set. The probability density function is determined based on the normal distribution characteristics of wind power prediction errors, and the upper and lower limits of integration are the upper and lower boundary values ​​of the corresponding confidence interval. For each confidence level set, the mean square error between the predicted wind power value and the historical measured value within the same confidence interval is extracted, and the ratio of this mean square error to the half-width of the corresponding confidence interval is calculated to obtain the probability deviation parameter for that confidence level set.

[0049] Then, using the sum of probability mass for each confidence level set as the numerator and the probability deviation parameter as the denominator, the initial weight value for that confidence level is calculated. The three confidence levels are then numbered, with k=1, k=2, and k=3 corresponding to the confidence level sets at 99%, 95%, and 90% confidence levels, respectively. The formula for calculating the initial weight value is as follows: ,in Let be the initial weight value for the k-th confidence level. Let be the sum of probability quality for the k-th confidence level set. Let be the probability deviation parameter for the k-th confidence level set. Then, divide the initial weight values ​​of the three confidence levels by the sum of the three initial weight values ​​respectively to normalize the weights, thus constructing the normalized weight function. The complete expression of this function is: ,in Let be the final probability weight factor corresponding to the k-th confidence level, and let be the input variable of the function, which is the initial weight value of each confidence level. , , The function outputs normalized weight values ​​in the interval from 0 to 1, and satisfies... The constraints.

[0050] Finally, the initial weight values ​​corresponding to the three confidence levels are substituted into the normalized weight function described above to calculate the normalized weight values ​​corresponding to the 99%, 95%, and 90% confidence levels. The final output is the multi-level probability weight factor for each confidence level, where the multi-level probability weight factor is the final weight value calculated by the normalized weight function described above for each confidence level. , , , This corresponds to a confidence level of 99%. The confidence level corresponds to a 95% confidence level. The confidence level corresponds to a 90% confidence level.

[0051] Furthermore, the method provided in this application embodiment includes:

[0052] The probability-modulated node uncertainty energy influence matrix is ​​multiplied by the inverse of the equivalent electrical distance between grid nodes, and the transfer gain function between node pairs is constructed by combining the power flow sensitivity partial derivative coefficient to generate the risk coupling strength coefficient matrix.

[0053] Specifically, firstly, the node admittance matrix of the target distribution circuit system is calculated using the node admittance matrix solution method of the power system. The node impedance matrix is ​​then obtained by inverting the node admittance matrix. Based on the node impedance matrix, the equivalent electrical distance between any two nodes in the power grid is calculated. The equivalent electrical distance is calculated using the ratio of mutual impedance to self-impedance in the node impedance matrix. The reciprocal of the calculated equivalent electrical distance for each node pair is taken to obtain the electrical distance influence coefficient for the corresponding node pair. Next, the node uncertainty energy influence matrix after probabilistic modulation is obtained. The row and column dimensions of this matrix are consistent with the total number of nodes in the power grid. Each element in the matrix corresponds to the initial amplitude of the uncertainty energy influence between a specific node pair. The node uncertainty energy influence matrix and the electrical distance influence coefficients of each node pair are then multiplied element-wise to obtain the initial coupling amplitude matrix of the node pair.

[0054] Next, combining the power flow sensitivity partial derivative coefficients calculated by the Newton-Raphson method in the previous steps, a transfer gain function between node pairs is constructed. The output value of the transfer gain function is the product of the initial coupling amplitude of the node pair and the corresponding power flow sensitivity partial derivative coefficient of the node pair. The output values ​​of the transfer gain functions of all node pairs in the power grid are integrated in the order of node number to generate a two-dimensional risk coupling strength coefficient matrix. The rows of the matrix correspond to energy outflow nodes, and the columns correspond to energy inflow nodes. Each element in the matrix represents the risk coupling strength coefficient between the corresponding two nodes. The larger the coefficient value, the stronger the correlation between the wind power uncertainty energy transfer capability and risk coupling between the nodes.

[0055] Furthermore, the method provided in this application embodiment includes:

[0056] At each probability level, the risk coupling strength coefficient matrix is ​​decomposed into eigenvalues, and the eigenvector corresponding to the largest eigenvalue is extracted as the main diffusion direction vector of uncertain energy in the probability scenario. The seepage direction parameter is determined according to the sign of the eigenvector.

[0057] Specifically, firstly, according to the probability levels corresponding to three confidence levels of 99%, 95%, and 90%, coefficient sub-matrices are extracted from the risk coupling strength coefficient matrix for each probability level. Each sub-matrix fully represents the risk coupling relationship between nodes in the power grid under a single probability scenario. Using the QR decomposition method from the linear algebra domain, eigenvalue decomposition is performed on the coefficient sub-matrix under each probability level. During the decomposition, the coefficient sub-matrix is ​​decomposed into the product of an eigenvalue diagonal matrix and an eigenvector matrix. After decomposition, all eigenvalues ​​are sorted in descending order of value. The eigenvector corresponding to the largest eigenvalue after sorting is extracted, and this eigenvector is determined as the main diffusion direction vector of uncertain energy under the corresponding probability scenario.

[0058] Next, the seepage direction parameters of the corresponding nodes are determined according to the sign attribute of each element in the feature vector. When the element sign is positive, it means that the uncertain energy flows out from the node corresponding to the element. When the element sign is negative, it means that the uncertain energy flows into the node corresponding to the element. Combining the numerical amplitude and sign attribute of the feature vector elements, the quantitative determination of the seepage direction parameters of uncertain energy between nodes under different probability scenarios is completed, forming a set of uncertain energy seepage direction parameters for multiple scenarios.

[0059] Furthermore, the method provided in this application embodiment includes:

[0060] In the risk seepage topology graph, the node seepage intensity index is calculated based on the risk coupling strength coefficient and seepage direction parameters corresponding to each edge weight. The node seepage intensity index is a weighted combination of the sum of the weights of the incoming edges and the sum of the weights of the outgoing edges. Path cumulative gain operation is performed on the risk seepage topology graph to calculate the cumulative coupling strength value on each directed path, and directed paths with cumulative values ​​exceeding a preset seepage threshold are configured as high-density channels of uncertain energy. Based on the node seepage intensity index and path cumulative gain results, the node flow concentration index is calculated, and key coupling nodes are identified based on the node flow concentration index. Under different probability level scenarios, the corresponding high-density channels of uncertain energy and key coupling nodes are identified respectively, and intersection enhancement and union expansion analysis are performed under each scenario to construct a set of risk seepage channels.

[0061] Specifically, firstly, a directed graph adjacency list traversal method from graph theory is used to traverse all power grid nodes in the multi-probability hierarchical weighted directed risk seepage topology graph, with each power grid node uniquely corresponding to a vertex in the topology graph. Combining the seepage direction parameters of each directed edge in the topology graph, incoming and outgoing edges are distinguished for each node, where directed edges pointing to the node are incoming edges, and directed edges extending outward from the node are outgoing edges. The risk coupling strength coefficients corresponding to all incoming edges of each node are calculated, and the sum of the weights of the incoming edges of the node is obtained by arithmetic summation. Simultaneously, the risk coupling strength coefficients corresponding to all outgoing edges of the node are calculated, and the sum of the weights of the outgoing edges of the node is obtained by arithmetic summation. A weighted combination method is used to calculate the node seepage strength index, with the weight of the sum of the weights of incoming edges set to 0.6 and the weight of the sum of the weights of outgoing edges set to 0.4. The weighted sum is then normalized using a min-max method, mapping it to the interval between 0 and 1, finally obtaining the node seepage strength index corresponding to each power grid node.

[0062] Next, a depth-first traversal algorithm from graph theory is used to perform a full path traversal of the risk seepage topology graph. The traversal process starts at the nodes where the wind farm connects to the grid and ends at the load nodes in the grid. Only directed paths that satisfy the seepage direction parameter constraints are traversed. Specifically, the seepage direction parameter constraints are based on the uncertain energy main diffusion direction vector corresponding to the largest eigenvalue after eigenvalue decomposition of the risk coupling strength coefficient matrix at each probability level. Rigid constraints on energy inflow and outflow between nodes are determined according to the signs of the vector elements. For any directed edge formed by two adjacent nodes in the grid, a directed edge only satisfies the seepage direction parameter constraints and can be included in the path traversal range if the element sign of the main diffusion direction vector corresponding to the starting node of the directed edge is positive (representing uncertain energy flowing out of the node) and the element sign of the main diffusion direction vector corresponding to the ending node is negative (representing uncertain energy flowing into the node), and the direction of the directed edge completely matches the main diffusion direction of uncertain energy. Simultaneously, it is required that no duplicate nodes appear within a single path to avoid forming closed-loop paths.

[0063] Subsequently, for each valid directed path obtained through traversal, a path cumulative gain operation is performed. The operation proceeds along the directed edges of the path, sequentially accumulating the risk coupling strength coefficients corresponding to each directed edge to obtain the cumulative coupling strength value for that directed path. Then, a preset seepage threshold is set using the quantile method. The cumulative coupling strength values ​​of all valid directed paths are sorted in descending order, and the values ​​corresponding to the top 20% quantiles are used as the preset seepage threshold. After setting the threshold, directed paths with cumulative coupling strength values ​​exceeding the preset seepage threshold are configured as high-density channels for uncertain energy, completing the initial identification of high-density channels in a single scenario.

[0064] Based on the node seepage intensity index and path cumulative gain calculated in the above steps, the node flow concentration index corresponding to each grid node is calculated. Specifically: First, all high-density channels of uncertain energy flowing through the corresponding node are counted. The cumulative coupling strength values ​​of these channels are arithmetically summed to obtain the cumulative risk flow value of the node. Then, the sum of the cumulative coupling strength values ​​of all high-density channels of uncertain energy in the entire topology is calculated to obtain the total risk flow value of the entire network. The node flow concentration index is calculated as the ratio of the cumulative risk flow value of the corresponding node to the total risk flow value of the entire network. This index can directly quantify the core position of the node in the wind power uncertainty risk propagation process. The node flow concentration indices of all grid nodes are sorted in descending order, and the top 10% of nodes after sorting are identified as key coupling nodes for grid risk propagation.

[0065] Subsequently, based on the probability levels corresponding to 99%, 95%, and 90% confidence levels, three independent probability analysis scenarios were defined. Under each probability scenario, the entire process of calculating the node seepage intensity index, identifying high-density channels with uncertain energy, and identifying key coupling nodes was executed, resulting in independent sets of high-density channels and key coupling nodes for each probability scenario. On one hand, intersection enhancement analysis was performed on the analysis results of the three probability scenarios to extract common elements from the three high-density channel sets. High-density channels appearing simultaneously in all three probability scenarios were marked as core risk channels. Similarly, common elements from the three key coupling node sets were extracted, and nodes appearing simultaneously in all three probability scenarios were marked as core key nodes. On the other hand, union expansion analysis was performed on the analysis results of the three probability scenarios to integrate all elements from the three high-density channel sets, supplementing scenario-based extended risk channels beyond the core risk channels. Similarly, all elements from the three key coupling node sets were integrated, supplementing scenario-based key nodes beyond the core key nodes.

[0066] Finally, the core risk channels, scenario-based extended risk channels, core key nodes, and scenario-based key nodes are structurally integrated to form a complete set of risk seepage channels. This set synchronously stores the probability level attributes, risk coupling strength values, and seepage direction parameters corresponding to each channel and node.

[0067] Through a series of steps including directed graph traversal, path cumulative gain calculation, node feature quantification, and multi-probability scenario set operation, the core path and key nodes of wind power uncertainty propagation in the power grid were accurately identified, providing accurate and comprehensive risk data support for the subsequent construction of resilient risk envelope and the formulation of power grid energy management strategies.

[0068] Furthermore, the method provided in this application embodiment includes:

[0069] In the set of risk seepage channels, the temporal rate of change modulus and spatial diffusion coefficient of the uncertainty energy evolution tensor corresponding to each channel are extracted, and a channel risk evolution index is constructed by combining the cumulative coupling strength of the corresponding channel path. A dynamic threshold function is constructed based on the channel risk evolution index, which is a weighted combination function of the temporal rate of change and the spatial diffusion coefficient. The comprehensive risk value of each node in the risk seepage topology is calculated, and the comprehensive risk value of the node is compared with the output value of the dynamic threshold function. The risk envelope boundary is configured according to the comparison result. During the continuous prediction period, when the increment of the channel risk evolution index exceeds the preset change ratio, the risk envelope boundary is expanded or contracted to form an elastic risk envelope region with adaptively adjustable boundaries.

[0070] In this embodiment, the cumulative coupling strength of the path is the cumulative value of the coupling strength obtained by sequentially adding the risk coupling strength coefficients of each directed edge along the seepage direction on a certain directed path in the risk seepage topology graph. It is used to quantify the overall transmission coupling degree of wind power uncertain energy on a single directed path.

[0071] In one embodiment, for each high-density channel within the risk seepage channel set, the time gradient component of the corresponding uncertain energy evolution tensor is extracted. The first-order backward difference method is used to calculate the time change rate of wind power within the continuous prediction period. The absolute value of the calculated time change rate is taken to obtain the time change rate magnitude. The larger the time change rate magnitude, the more drastic the time dimension change of wind power fluctuations, and the stronger the risk dynamic evolution characteristics of the corresponding channel. Simultaneously, the spatial diffusion component of the corresponding uncertain energy evolution tensor is extracted. The normalized value of the spatial diffusion component is used as the spatial diffusion coefficient corresponding to the channel. The larger the spatial diffusion coefficient, the stronger the spatial transmission capability of wind power fluctuations between grid nodes, and the wider the risk diffusion range of the corresponding channel. A channel risk evolution index is constructed using a linear weighted combination method. The three parameters—temporal rate of change modulus, spatial diffusion coefficient, and path cumulative coupling strength—are each normalized using a min-max method and mapped to the interval between 0 and 1. Weight coefficients of 0.3, 0.3, and 0.4 are set for the three parameters respectively. The normalized parameters are multiplied by their corresponding weight coefficients and then summed to obtain the channel risk evolution index for a single channel. This index can fully quantify the comprehensive risk level of a single high-density channel in the three dimensions of time, space, and probability.

[0072] Next, based on the calculated risk evolution indices for each channel, a dynamic threshold function is constructed. This function is a weighted combination of the time rate of change and the spatial diffusion coefficient, corresponding to the calculation logic of the channel risk evolution index. First, global normalization is performed on the temporal rate of change magnitude and spatial diffusion coefficient for all channels to eliminate dimensional differences between different parameters. A weighting coefficient of 0.5 is set for both the temporal rate of change magnitude and the spatial diffusion coefficient, constructing a basic weighting function. Then, the average value of the channel risk evolution index for all channels is used as a correction coefficient and introduced into the basic weighting function to complete the final construction of the dynamic threshold function. The calculation logic of the dynamic threshold function is as follows: the dynamic threshold output value equals 0.5 multiplied by the globally normalized temporal rate of change magnitude, plus 0.5 multiplied by the globally normalized spatial diffusion coefficient, and then the arithmetic mean of the risk evolution indices for all channels. The function takes as input the temporal rate of change modulus and spatial diffusion coefficient for the corresponding prediction period, and outputs a dynamic threshold that changes in real time with the characteristics of wind power fluctuations. This threshold increases or decreases synchronously with the severity of wind power fluctuations, adapting to the changes in risk levels for different prediction periods.

[0073] For all grid nodes in the risk seepage topology, the comprehensive risk value of each node is calculated. This calculation combines the node seepage intensity index, node flow concentration index, and the average channel risk evolution index of all high-density channels flowing through that node, obtained from the previous steps. The three parameters—node seepage intensity index, node flow concentration index, and average channel risk evolution index—are normalized with weighting coefficients of 0.3, 0.3, and 0.4, respectively. A weighted summation is then performed to obtain the comprehensive risk value for each node. This value comprehensively quantifies the degree of risk to a single grid node due to wind power uncertainty. The comprehensive risk values ​​of all nodes are compared point-by-point with the output value of the dynamic threshold function for the current prediction period. If a node's comprehensive risk value is greater than or equal to the dynamic threshold output value, the node is marked as being within the risk envelope; if its comprehensive risk value is less than the dynamic threshold output value, the node is marked as being outside the risk envelope. Taking all nodes within the risk envelope as the core, traverse their adjacent nodes to determine the boundary nodes of the risk envelope, and connect the boundary nodes in sequence to form a closed risk envelope boundary, thus fully defining the core impact range of wind power uncertainty risk.

[0074] Finally, within a continuous prediction period, the channel risk evolution index of each channel is recalculated periodically according to a fixed prediction step size. The channel risk evolution index of the current prediction period is compared with that of the previous prediction period, and the incremental value and incremental ratio of the risk evolution index of each channel are calculated. The incremental ratio is the ratio of the incremental value to the channel risk evolution index of the previous prediction period. A preset change ratio is set to 20%. When the incremental ratio of the average channel risk evolution index of all channels exceeds the preset change ratio, it is determined that the power grid risk level has increased significantly. At this time, the output value of the dynamic threshold function and the comprehensive risk value of each node are recalculated, the coverage of nodes within the risk envelope is increased, and the risk envelope boundary is updated outward. When the decrease ratio of the average channel risk evolution index of all channels exceeds the preset change ratio, it is determined that the power grid risk level has decreased significantly. At this time, the output value of the dynamic threshold function and the comprehensive risk value of each node are recalculated, the coverage of nodes within the risk envelope is reduced, and the risk envelope boundary is updated inward. When the change ratio of the channel risk evolution index does not exceed the preset change ratio, the current risk envelope boundary remains unchanged. By dynamically updating the boundaries during the continuous forecast period, an elastic risk envelope area is ultimately formed that can adaptively adjust the boundaries according to the level of wind power uncertainty risk. This envelope area fully covers the core impact nodes and risk propagation paths of wind power uncertainty.

[0075] Through a series of steps including channel risk evolution index quantification, dynamic threshold function construction, node comprehensive risk assessment, and boundary adaptive updating, the scope of wind power uncertainty risk impact is accurately defined and dynamically adapted, providing a spatial boundary basis that can be adjusted in real time according to the risk level for subsequent grid-level energy regulation and allocation.

[0076] Furthermore, the method provided in this application embodiment includes:

[0077] Within the elastic risk envelope, grid nodes are divided into high-risk, medium-risk, and low-risk zones, and risk weight coefficients are determined based on the number of nodes covered by each risk zone and the risk evolution index. An energy storage availability matrix is ​​constructed based on the stored grid energy, and the energy storage availability matrix is ​​coupled with the risk weight coefficients to calculate the target energy allocation for each risk zone. Based on the target energy allocation, a hierarchical energy regulation and allocation under high-medium-low risk constraints is performed to establish a grid energy management strategy.

[0078] Optionally, firstly, the comprehensive risk values ​​of all power grid nodes within the elastic risk envelope are extracted. The average frequency quantile method is used to classify the comprehensive risk values ​​of the nodes, classifying the top 20% of nodes by comprehensive risk value as high-risk areas, the top 20% to 60% as medium-risk areas, and the bottom 40% as low-risk areas. The number of power grid nodes covered in each risk area is counted, and the arithmetic mean of the channel risk evolution index of all nodes in each risk area is calculated to obtain the regional average risk evolution index for that risk area. A linear weighted combination method is used to construct a risk weight coefficient calculation model, setting a weight coefficient of 0.4 for the number of nodes covered in the risk area and a weight coefficient of 0.6 for the regional average risk evolution index. Weighted summation is performed on the high, medium, and low risk areas respectively to obtain the initial risk weight value for each risk area. Then, the three initial risk weight values ​​are normalized so that the sum of the three risk weight coefficients is 1, finally yielding the risk weight coefficients corresponding to the high, medium, and low risk areas.

[0079] Then, real-time operating data of all energy storage power stations within the power grid are acquired, including the rated capacity, current state of charge, maximum charging and discharging power, and available charging and discharging duration of each power station. The maximum available energy value that each power station can access during the current forecast period is calculated. This value is the product of the power station's rated capacity and current state of charge, multiplied by a preset available capacity coefficient, which is fixed at 0.8. Using the power stations within the power grid as the row dimension and the continuous forecast period as the column dimension, the maximum available energy value of each power station during the corresponding forecast period is filled into the corresponding position in the matrix, constructing a two-dimensional energy storage available energy matrix. The summation calculation is performed on all elements in the energy storage available energy matrix to obtain the total available energy of the power grid currently available. The total available energy of energy storage is multiplied by the risk weight coefficients corresponding to high, medium, and low risk zones respectively to calculate the target energy allocation for each risk zone during the current forecast period. The target energy allocation is positively correlated with the risk weight coefficient of the corresponding risk zone, with the highest target energy value allocated to the high-risk zone, followed by the medium-risk zone, and the lowest to the low-risk zone.

[0080] Finally, a tiered energy regulation and allocation system is implemented, prioritizing high-risk areas, followed by medium-risk areas, and lastly low-risk areas. For high-risk areas, the full amount of available energy storage is utilized according to the target energy allocation, employing a constant-power charge-discharge mode to smooth power fluctuations and mitigate node power deviations caused by wind power fluctuations. This smoothing process prioritizes key coupling nodes and the start and end nodes of uncertain high-density energy channels within the high-risk area. For medium-risk areas, 80% of the available energy storage is utilized, employing a follow-up charge-discharge mode to smooth power fluctuations, covering major distribution nodes and risk channel-related nodes within the medium-risk area. For low-risk areas, 50% of the available energy storage is utilized, employing a standby regulation mode to correct power deviations, reserving the remaining available energy as emergency grid regulation capacity. During the continuous forecast period, the risk zone division results, risk weight coefficients, available energy storage matrix, and target energy allocation of the flexible risk envelope are updated periodically, and the energy regulation and allocation schemes for each risk zone are adjusted synchronously, forming a closed-loop grid energy management strategy covering all time periods.

[0081] In summary, the wind power forecasting-driven grid energy management method provided in this application has the following technical effects:

[0082] This application constructs an uncertain energy evolution tensor by acquiring wind farm power prediction distribution sequences and confidence interval data, maps it to the grid node topology to generate a risk seepage topology map, identifies high-density channels and key nodes, constructs an elastic risk envelope region, and combines grid energy storage to execute hierarchical energy regulation and allocation, establishing a wind power prediction-driven grid energy management strategy. This improves the accuracy and adaptability of grid energy management, achieving the technical effect of accurate risk identification and dynamic control within the grid, and completing differentiated energy regulation and allocation in conjunction with grid energy storage.

[0083] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a wind power forecast-driven grid energy management system, the system comprising:

[0084] Energy evolution tensor construction module 1 is used to obtain the power prediction distribution sequence and corresponding confidence interval data of the wind farm in multiple prediction periods, and construct an uncertain energy evolution tensor containing time gradient component, spatial diffusion component and probability discrete component based on the obtained results.

[0085] Risk seepage topology acquisition module 2 is used to map the uncertainty energy evolution tensor to the topology of the power grid nodes, calculate the risk coupling strength coefficient and uncertainty energy seepage direction parameters of each power grid node, and generate a risk seepage topology map that characterizes the propagation trend of uncertainty in the power grid.

[0086] Risk seepage channel set construction module 3, which identifies high-density channels with uncertain energy and key coupling nodes based on the risk seepage topology map, and constructs a risk seepage channel set.

[0087] The elastic risk envelope acquisition module 4 is used to calculate the partition elastic threshold on the risk seepage channel set based on the time change rate of the uncertain energy evolution tensor and the spatial diffusion coefficient, and generate an elastic risk envelope with adaptively adjustable boundaries.

[0088] The power grid energy management strategy construction module 5 is used to perform hierarchical energy regulation and allocation based on the stored power grid energy and the elastic risk envelope area to establish a power grid energy management strategy.

[0089] Furthermore, the risk seepage topology map acquisition module 2 is used to perform the following steps:

[0090] The time gradient component in the uncertain energy evolution tensor is projected onto the power injection change of the grid nodes. The spatial diffusion component is coupled with the grid branch impedance matrix and power flow sensitivity matrix to construct the node uncertain energy influence matrix. The confidence interval hierarchical discretization processing is performed on the probability discrete component to obtain multi-level probability weighting factors. These multi-level probability weighting factors are introduced into the node uncertain energy influence matrix to weight and correct the uncertainty energy influence amplitude of each node, generating a probability-modulated node uncertain energy influence matrix. Based on the probability-modulated node uncertain energy influence matrix and the electrical distance coefficient between grid nodes, the uncertainty energy transfer gain between nodes is calculated to form a risk coupling strength coefficient matrix. Conditional gradient divergence is performed on the risk coupling strength coefficient matrix at different probability levels to extract the main diffusion direction vector of uncertain energy under different probability scenarios, thereby determining the uncertain energy seepage direction parameters in multiple scenarios. Using the risk coupling strength coefficient as the edge weight and the uncertain energy seepage direction in multiple scenarios as the directed constraint, a multi-probability level weighted directed risk seepage topology is constructed.

[0091] Furthermore, the risk seepage topology map acquisition module 2 is used to perform the following steps:

[0092] Based on the confidence level intervals corresponding to each discrete probability node in the discrete probability component, the discrete probability nodes are divided into at least three confidence level sets; the sum of probability quality and probability deviation parameters within each confidence level set are statistically analyzed; a normalized weight function is constructed based on the sum of probability quality and probability deviation parameters of each confidence level set, and the corresponding multi-level probability weight factors are calculated.

[0093] Furthermore, the risk seepage topology map acquisition module 2 is used to perform the following steps:

[0094] The probability-modulated node uncertainty energy influence matrix is ​​multiplied by the inverse of the equivalent electrical distance between grid nodes, and the transfer gain function between node pairs is constructed by combining the power flow sensitivity partial derivative coefficient to generate the risk coupling strength coefficient matrix.

[0095] Furthermore, the risk seepage topology map acquisition module 2 is used to perform the following steps:

[0096] At each probability level, the risk coupling strength coefficient matrix is ​​decomposed into eigenvalues, and the eigenvector corresponding to the largest eigenvalue is extracted as the main diffusion direction vector of uncertain energy in the probability scenario. The seepage direction parameter is determined according to the sign of the eigenvector.

[0097] Furthermore, the risk seepage channel assembly construction module 3 is used to perform the following steps:

[0098] In the risk seepage topology graph, the node seepage intensity index is calculated based on the risk coupling strength coefficient and seepage direction parameters corresponding to each edge weight. The node seepage intensity index is a weighted combination of the sum of the weights of the incoming edges and the sum of the weights of the outgoing edges. Path cumulative gain operation is performed on the risk seepage topology graph to calculate the cumulative coupling strength value on each directed path, and directed paths with cumulative values ​​exceeding a preset seepage threshold are configured as high-density channels of uncertain energy. Based on the node seepage intensity index and path cumulative gain results, the node flow concentration index is calculated, and key coupling nodes are identified based on the node flow concentration index. Under different probability level scenarios, the corresponding high-density channels of uncertain energy and key coupling nodes are identified respectively, and intersection enhancement and union expansion analysis are performed under each scenario to construct a set of risk seepage channels.

[0099] Furthermore, the elastic risk envelope acquisition module 4 is used to perform the following steps:

[0100] In the set of risk seepage channels, the temporal rate of change modulus and spatial diffusion coefficient of the uncertainty energy evolution tensor corresponding to each channel are extracted, and a channel risk evolution index is constructed by combining the cumulative coupling strength of the corresponding channel path. A dynamic threshold function is constructed based on the channel risk evolution index, which is a weighted combination function of the temporal rate of change and the spatial diffusion coefficient. The comprehensive risk value of each node in the risk seepage topology is calculated, and the comprehensive risk value of the node is compared with the output value of the dynamic threshold function. The risk envelope boundary is configured according to the comparison result. During the continuous prediction period, when the increment of the channel risk evolution index exceeds the preset change ratio, the risk envelope boundary is expanded or contracted to form an elastic risk envelope region with adaptively adjustable boundaries.

[0101] Furthermore, the grid energy management strategy construction module 5 is used to perform the following steps:

[0102] Within the elastic risk envelope, grid nodes are divided into high-risk, medium-risk, and low-risk zones, and risk weight coefficients are determined based on the number of nodes covered by each risk zone and the risk evolution index. An energy storage availability matrix is ​​constructed based on the stored grid energy, and the energy storage availability matrix is ​​coupled with the risk weight coefficients to calculate the target energy allocation for each risk zone. Based on the target energy allocation, a hierarchical energy regulation and allocation under high-medium-low risk constraints is performed to establish a grid energy management strategy.

[0103] Example 3, as Figure 3 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides an electronic device, the electronic device comprising:

[0104] The memory 303 is used to store executable instructions; the processor 302 is used to implement the wind power prediction-driven grid energy management method when executing the executable instructions stored in the memory 303.

[0105] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention. This electronic device is in the form of a general-purpose computing device, and its components may include, but are not limited to, an input device 301, a processor 302, a memory 303, and an output device 304. The processor 302 may be one or more; the memory 303 may include a computer-readable medium and at least one program product having a set (at least one) of program modules configured to perform the functions of the embodiments of this application.

[0106] The memory 303 shown in the embodiments of the present invention can be any combination of one or more computer-readable media; the computer-readable storage media can be, but is not limited to, infrared, semiconductor systems, devices or components, or any combination thereof, for storing software programs, computer-executable programs and modules, such as the program instructions / modules corresponding to the wind power forecast-driven grid energy management method in the embodiments of the present invention. The processor 302 executes various functional applications and data processing of the computer device by running the software programs, instructions and modules stored in the memory 303, thereby realizing the above-mentioned wind power forecast-driven grid energy management method.

[0107] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method of grid energy management driven by wind power prediction, characterized in that, The method includes: Obtain the power prediction distribution sequence and corresponding confidence interval data of wind farms in multiple prediction periods, and construct an uncertain energy evolution tensor containing temporal gradient components, spatial diffusion components and probability discrete components based on the obtained results; The uncertainty energy evolution tensor is mapped to the grid node topology, and the risk coupling strength coefficient and uncertainty energy seepage direction parameters of each grid node are calculated to generate a risk seepage topology map that characterizes the propagation trend of uncertainty in the grid. Based on the aforementioned risk seepage topology map, identify high-density channels with uncertain energy and key coupling nodes, and construct a set of risk seepage channels; Based on the time change rate of the uncertain energy evolution tensor and the spatial diffusion coefficient, the partition elastic threshold is calculated on the set of risk seepage channels to generate an elastic risk envelope region with adaptively adjustable boundaries. A grid energy management strategy is established by performing hierarchical energy regulation and allocation based on the stored grid energy and the elastic risk envelope. Generate a risk seepage topology map that characterizes the propagation trend of uncertainty in the power grid, including: The time gradient component in the uncertain energy evolution tensor is projected onto the power injection change of the grid node, and the spatial diffusion component is coupled with the grid branch impedance matrix and power flow sensitivity matrix to construct the node uncertain energy influence matrix. The confidence interval hierarchical discretization process is performed on the probability discrete components to obtain multi-level probability weighting factors. The multi-level probability weighting factors are then introduced into the node uncertainty energy influence matrix. The uncertainty energy influence amplitude of each node is weighted and corrected to generate the probability-modulated node uncertainty energy influence matrix. Based on the probability-modulated node uncertainty energy influence matrix and the electrical distance coefficient between power grid nodes, the uncertainty energy transfer gain between nodes is calculated to form the risk coupling strength coefficient matrix. Based on the risk coupling strength coefficient matrix, conditional gradient divergence operation is performed at different probability levels to extract the main diffusion direction vector of uncertain energy under different probability scenarios, so as to determine the direction parameters of uncertain energy seepage in multiple scenarios. Using the risk coupling strength coefficient as the edge weight and the multi-scenario uncertain energy seepage direction as the directed constraint, a multi-probability hierarchical weighted directed risk seepage topology is constructed. Based on the aforementioned risk seepage topology map, high-density channels with uncertain energy and key coupling nodes are identified, and a set of risk seepage channels is constructed, including: In the risk seepage topology diagram, the node seepage intensity index is calculated based on the risk coupling strength coefficient and seepage direction parameters corresponding to each edge weight. The node seepage intensity index is a weighted combination of the sum of the weights of the incoming edges and the sum of the weights of the outgoing edges of the node. Perform path cumulative gain calculation on the risk seepage topology map, calculate the cumulative value of coupling strength on each directed path, and configure the directed path with the cumulative value exceeding the preset seepage threshold as a high-density channel of uncertain energy. Based on the node seepage intensity index and path cumulative gain results, the node flow concentration index is calculated, and key coupled nodes are identified according to the node flow concentration index. Under different probability levels, the corresponding high-density channels of uncertain energy and key coupling nodes are identified, and intersection enhancement and union expansion analysis are performed under each scenario to construct a set of risk seepage channels.

2. The wind power forecast-driven grid energy management method as described in claim 1, characterized in that, Perform confidence interval hierarchical discretization on the discrete probability components, including: Based on the confidence level intervals corresponding to each discrete probability node in the discrete probability component, the discrete probability nodes are divided into at least three confidence level sets. Calculate the sum of probability quality and probability deviation parameters within each confidence level set; Based on the sum of probability quality and probability deviation parameters of each confidence level set, a normalized weight function is constructed, and the corresponding multi-level probability weight factor is calculated.

3. The wind power forecast-driven grid energy management method as described in claim 1, characterized in that, Based on the probability-modulated node uncertainty energy influence matrix and the electrical distance coefficient between grid nodes, the inter-node uncertainty energy transfer gain is calculated to form a risk coupling strength coefficient matrix, including: The probability-modulated node uncertainty energy influence matrix is ​​multiplied by the inverse of the equivalent electrical distance between grid nodes, and the transfer gain function between node pairs is constructed by combining the power flow sensitivity partial derivative coefficient to generate the risk coupling strength coefficient matrix.

4. The wind power forecast-driven grid energy management method as described in claim 1, characterized in that, Perform conditional gradient divergence calculations based on the risk coupling strength coefficient matrix at different probability levels, including: At each probability level, the risk coupling strength coefficient matrix is ​​decomposed into eigenvalues, and the eigenvector corresponding to the largest eigenvalue is extracted as the main diffusion direction vector of uncertain energy in the probability scenario. The seepage direction parameter is determined according to the sign of the eigenvector.

5. The wind power forecast-driven grid energy management method as described in claim 1, characterized in that, Generate a resilient risk envelope region with adaptively adjustable boundaries, including: In the set of risk seepage channels, the temporal rate of change modulus and spatial diffusion coefficient of the uncertainty energy evolution tensor corresponding to each channel are extracted, and the channel risk evolution index is constructed by combining the cumulative coupling strength of the path of the corresponding channel. A dynamic threshold function is constructed based on the channel risk evolution index. The dynamic threshold function is a weighted combination function of the time change rate and the spatial diffusion coefficient. Calculate the comprehensive risk value of each node in the risk seepage topology diagram, compare the comprehensive risk value of the node with the output value of the dynamic threshold function, and configure the risk envelope boundary according to the comparison result; During the continuous prediction period, when the increment of the channel risk evolution index exceeds the preset change ratio, the risk envelope boundary is expanded or contracted to form an elastic risk envelope area with adaptively adjustable boundaries.

6. The wind power forecast-driven grid energy management method as described in claim 1, characterized in that, Based on the stored grid energy and the aforementioned resilient risk envelope, a hierarchical energy regulation and allocation strategy is established to create a grid energy management strategy, including: Within the flexible risk envelope, power grid nodes are divided into high-risk, medium-risk, and low-risk zones, and risk weight coefficients are determined based on the number of nodes covered by each risk zone and the risk evolution index. Based on the stored grid energy, an energy storage available energy matrix is ​​constructed, and the energy storage available energy matrix is ​​coupled with the risk weight coefficient to calculate the target energy allocation for each risk zone; Based on the target energy allocation, a tiered energy regulation and allocation under high-medium-low risk constraints is implemented to establish a power grid energy management strategy.

7. A wind power forecast-driven grid energy management system, characterized in that, For implementing the wind power prediction-driven grid energy management method according to any one of claims 1-6, the system comprises: The energy evolution tensor construction module is used to obtain the power prediction distribution sequence and corresponding confidence interval data of wind farms in multiple prediction periods, and construct an uncertain energy evolution tensor containing time gradient components, spatial diffusion components and probability discrete components based on the obtained results. The risk seepage topology acquisition module is used to map the uncertainty energy evolution tensor to the grid node topology, calculate the risk coupling strength coefficient and uncertainty energy seepage direction parameters of each grid node, and generate a risk seepage topology map that characterizes the propagation trend of uncertainty in the grid. The risk seepage channel set construction module identifies high-density channels with uncertain energy and key coupling nodes based on the risk seepage topology map, and constructs a risk seepage channel set. The elastic risk envelope acquisition module is used to calculate the partition elastic threshold on the risk seepage channel set based on the time change rate of the uncertain energy evolution tensor and the spatial diffusion coefficient, and generate an elastic risk envelope with adaptively adjustable boundaries. The power grid energy management strategy construction module is used to perform hierarchical energy regulation and allocation based on the stored power grid energy and the elastic risk envelope area, and to establish a power grid energy management strategy.

8. An electronic device, characterized in that, The electronic device includes: Memory, used to store executable instructions; The processor, when executing executable instructions stored in the memory, implements the wind power forecast-driven grid energy management method according to any one of claims 1 to 6.

Images