Power network active defense method and system in open converged network environment

Abstract

The invention relates to the technical field of network security, in particular to a power network active defense method and system in an open fusion network environment, and the method comprises the steps: recognizing an electromagnetic field intensity gradient region according to the electromagnetic field distribution data among a plurality of groups of valve towers in a valve hall of an extra-high voltage converter station; generating an electromagnetic interference relation topological graph according to the electromagnetic field coupling strength among each group of valve towers in the electromagnetic field intensity gradient region, and obtaining a co-evolution relation between port link quality and an electromagnetic field topological structure by combining port link quality data of communication equipment near the electromagnetic field intensity gradient region; and according to the co-evolution relationship, analyzing the potential influence of the dynamic migration of the gradient region of the electromagnetic field intensity on the communication security, obtaining a communication security risk prediction result, and generating an optimal power network protection strategy through a reinforcement learning algorithm. According to the method, the influence of dynamic migration of an electromagnetic field intensity gradient region on communication safety is analyzed by monitoring electromagnetic field distribution in an extra-high voltage converter station valve hall, and active defense of a power network in an open fusion environment is realized.

Description

Technical Field

[0001] The present invention relates to the field of network security technology, and in particular to an active defense method and system for a power network in an open and integrated network environment. Background Art

[0002] In an open and integrated network environment, the stable operation of the power grid faces unprecedented challenges, and key power facilities such as UHV converter stations face more complex electromagnetic interference and network security threats, especially in the key area of the UHV converter station valve hall. The electromagnetic field distribution inside the valve hall is complex, and the electromagnetic field coupling between multiple groups of valve towers not only has variable intensity, but also its spatial distribution shows significant dynamic migration characteristics over time. This dynamic migration causes the position of the electromagnetic field intensity gradient zone to change, which seriously affects the communication link quality of the industrial control equipment in the valve hall. There is a spatiotemporal correlation between the link quality of the network port and the topological structure of the electromagnetic field. The two show complex dynamic behaviors of mutual influence and coordinated evolution, forming a dynamically balanced system.

[0003] However, this dynamic change of the electromagnetic field often leads to a sharp drop in the link signal-to-noise ratio, which in turn causes a significant increase in communication delay and packet loss rate. In extreme cases, it may even lead to the interruption of the communication link, posing a serious threat to the stable operation of the power network. At the same time, the fluctuation of link quality does not exist in isolation. It will produce feedback disturbances on the distribution of the electromagnetic field, further aggravating the dynamic correlation and evolutionary complexity between the electromagnetic field and the link. The current power network faces many difficulties in responding to this challenge. On the one hand, it is extremely difficult to obtain electromagnetic field measurement data, and there is a lack of in-depth understanding of the coordinated evolution of the electromagnetic field and link quality. On the other hand, the real-time perception accuracy of link quality also needs to be improved in order to more accurately reflect the impact of electromagnetic field changes on communication links. Due to these two limitations, the existing active defense mechanisms often lack the necessary trigger criteria and it is difficult to accurately identify security threats caused by electromagnetic distortion.

[0004] Therefore, how to break through the bottleneck of electromagnetic and link collaborative perception in an open and integrated network environment and deeply explore the evolution laws contained in cross-domain data has become a key issue that needs to be urgently solved in the current field of active defense of power networks to ensure the stable operation of power networks. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method and system for active defense of power network in an open and integrated network environment.

[0006] In a first aspect, the present invention provides a method for active defense of a power network in an open converged network environment, the method comprising the following steps: In an open and integrated network environment, analyze the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station, and identify the electromagnetic field intensity gradient change area; Analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area, and generate a topological map of electromagnetic interference relationships; Determine the migration direction of the electromagnetic field intensity gradient change area, and collect the port link quality data among communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area; According to the port link quality data and the topological map of electromagnetic interference relationships, obtain the co-evolution relationship between the port link quality and the electromagnetic field topological structure; Analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topological structure, and obtain the communication security risk prediction result; Based on the communication security risk prediction result, generate an optimal power network protection strategy through a reinforcement learning algorithm.

[0007] In a further implementation, the step of analyzing the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station and identifying the electromagnetic field intensity gradient change area includes: Divide the electromagnetic field distribution data among several groups of valve towers collected by the electromagnetic field sensor array in the valve hall of the UHV converter station according to different monitoring areas to obtain the electromagnetic field distribution data of each monitoring area; According to the electromagnetic field distribution data between adjacent monitoring areas, use spatial divergence to calculate the electromagnetic field intensity difference between different monitoring areas; Use a sliding time window to dynamically analyze the electromagnetic field intensity difference, and calculate the change rate of the electromagnetic field intensity gradient between regions; Determine the monitoring area where the change rate of the electromagnetic field intensity gradient between regions exceeds the preset electromagnetic field intensity gradient change threshold range as the electromagnetic field intensity gradient change area.

[0008] In a further implementation, the step of analyzing the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area and generating a topological map of electromagnetic interference relationships includes: Construct a vector electromagnetic field model of the valve tower array based on Maxwell's equations, and calculate the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area according to the vector electromagnetic field model; Introduce a shielding intensity parameter, and calculate the spatial attenuation coefficient of the electromagnetic field between each group of valve towers according to the electromagnetic field coupling intensity; According to the spatial attenuation coefficient, use the exponential attenuation function to fit the attenuation amount of the electromagnetic field strength at different propagation distances, and obtain the electromagnetic field propagation attenuation characteristic curve; Calculate the electromagnetic wave propagation path loss according to the electromagnetic field propagation attenuation characteristic curve, and use the Dijkstra algorithm to search for the shortest propagation path between each path node with the electromagnetic wave propagation path loss as the weight; Use the graph neural network to analyze the shortest propagation path and the electromagnetic field distribution data, calculate the topological correlation degree between each group of valve towers, and construct a topological graph of the electromagnetic interference relationship between the valve towers through the topological correlation degree.

[0009] In a further implementation, the step of determining the migration direction of the electromagnetic field strength gradient region includes: Based on the electromagnetic field distribution data, use a sliding time window to calculate the field strength mean value of the electromagnetic field strength gradient region within each time window, and obtain the field strength mean value sequence of the electromagnetic field strength gradient region; Calculate the standard deviation of the field strength within each time window according to the field strength mean value sequence of the electromagnetic field strength gradient region, and obtain the standard deviation time sequence; Perform wavelet analysis on the standard deviation time sequence, and extract the electromagnetic field strength change feature vector; Based on the electromagnetic field strength change feature vector, use the time series analysis method to model and predict the electromagnetic field strength change trend, and generate the electromagnetic field strength change trend matrix; Use the Kalman filter to perform real-time estimation on the electromagnetic field strength change trend matrix, and determine the migration direction of the electromagnetic field strength gradient region.

[0010] In a further implementation, the step of collecting the port link quality data between communication devices near the electromagnetic field strength gradient region according to the migration direction of the electromagnetic field strength gradient region includes: Collect the port data packet transmission rate between communication devices near the electromagnetic field strength gradient region based on the migration direction of the electromagnetic field strength gradient region; Calculate the port link bandwidth utilization rate according to the port data packet transmission rate and the port physical bandwidth; Obtain the port link quality data between communication devices near the electromagnetic field strength gradient region according to the port link bandwidth utilization rate and the pre-acquired link quality indicators.

[0011] In a further implementation, the step of obtaining the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topological graph includes: Analyze the port link quality data and the electromagnetic interference relationship topology graph using a graph mining method, and extract the time-varying feature sequences of the port link quality data and the electromagnetic interference relationship topology graph at different time points; Use the autoregressive integrated moving average method to perform modeling analysis on the time-varying feature sequences to obtain a topological evolution trend vector; Based on the topological evolution trend vector, use the dynamic mode decomposition algorithm to identify the co-evolution relationship between the port link quality and the electromagnetic field topological structure.

[0012] In a further implementation, the step of analyzing the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topological structure to obtain a communication security risk prediction result includes: Construct a co-evolution state transition matrix according to the co-evolution relationship between the port link quality and the electromagnetic field topological structure, and use a long short-term memory network to learn and predict the co-evolution state transition matrix to obtain the bandwidth limitation probability and the packet loss rate; Use the Markov method to perform state prediction on the bandwidth limitation probability and the packet loss rate to obtain the communication link interruption probability distribution; According to the characteristic parameters of the electromagnetic field intensity gradient region, use the spatio-temporal correlation analysis method to analyze the dynamic migration of the electromagnetic field intensity gradient region, and construct a spatial position determination matrix; Calculate the overlap degree between the communication link and the migration path of the gradient region according to the communication link interruption probability distribution and the spatial position determination matrix; If the overlap degree exceeds a preset overlap threshold, calculate the communication link interference probability according to the overlap degree, and generate a communication security risk prediction result according to the communication link interference probability.

[0013] In a further implementation, the step of analyzing the dynamic migration of the electromagnetic field intensity gradient region according to the characteristic parameters of the electromagnetic field intensity gradient region, using the spatio-temporal correlation analysis method, and constructing a spatial position determination matrix includes: According to the characteristic parameters of the electromagnetic field intensity gradient region, use Gaussian process regression to analyze the spatial distribution state of the electromagnetic field intensity gradient region at different time points to obtain the dynamic migration trajectory matrix of the electromagnetic field intensity gradient region; the characteristic parameters of the electromagnetic field intensity gradient region include the intensity gradient, the spatial distribution range, and the migration rate of the electromagnetic field intensity gradient region; Based on the dynamic migration trajectory matrix, determine the time point sequence when the magnetic field intensity gradient region changes and its position information at different time points; Using a spatio-temporal correlation analysis method to analyze the time point sequence of changes in the electromagnetic field intensity gradient region and its position information at different time points, and obtaining the device point density between communication devices near the electromagnetic field intensity gradient region and the link connectivity between communication devices; According to the device point density and the link connectivity, a spatial position determination matrix is constructed using a weighted algorithm.

[0014] In a further implementation, the step of generating an optimal power network protection strategy through a reinforcement learning algorithm based on the communication security risk prediction result includes: Combining the power network operating state and the communication security risk prediction result into a multi-dimensional state vector, and using the multi-dimensional state vector as the state space in the reinforcement learning algorithm; Defining protection strategies as the action space according to the protection requirements of the power network, and encoding each action in the action space to form a discrete action set; Initializing the Q-value table of the reinforcement learning algorithm. In the power network simulation environment under the open fusion network environment, the control agent selects and executes actions from the action space according to the current state, and iteratively updates the Q-value table according to the rewards obtained from the executed actions. Repeat the above process until the Q-value converges; According to the converged Q-value table, select the action with the maximum Q-value as the optimal protection strategy in the current state.

[0015] In a second aspect, the present invention provides a power network active defense system under an open fusion network environment, and the system includes: A gradient change recognition module, which is used to analyze the electromagnetic field space intensity gradient according to the electromagnetic field distribution data between several groups of valve towers in the valve hall of the UHV converter station under the open fusion network environment, and identify the electromagnetic field intensity gradient region; An electromagnetic analysis module, which is used to analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity between groups of valve towers in the electromagnetic field intensity gradient region, and generate an electromagnetic interference relationship topology diagram; A migration analysis module, which is used to determine the migration direction of the electromagnetic field intensity gradient region, and collect the port link quality data between communication devices near the electromagnetic field intensity gradient region according to the migration direction of the electromagnetic field intensity gradient region; A co-evolution module, which is used to obtain the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology diagram; A risk prediction module, which is used to analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtain a communication security risk prediction result; A policy generation module, configured to generate an optimal power network protection policy based on the communication security risk prediction result through a reinforcement learning algorithm.

[0016] The present invention provides a power network active defense method and system in an open and integrated network environment. The method analyzes the electromagnetic field space intensity gradient according to the electromagnetic field distribution data between several groups of valve towers in the valve hall of a UHV converter station, and identifies the electromagnetic field intensity gradient change area; analyzes the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity between the valve towers in the electromagnetic field intensity gradient change area, and generates an electromagnetic interference relationship topology map; collects the port link quality data between communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area; obtains the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology map; analyzes the potential impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtains a communication security risk prediction result; generates an optimal power network protection policy based on the communication security risk prediction result through a reinforcement learning algorithm. Compared with the prior art, this method accurately analyzes the impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security by monitoring the electromagnetic field distribution in the valve hall of the UHV converter station, and combines the electromagnetic interference relationship topology map and the co-evolution relationship to achieve the active defense of the power network in an open and integrated environment, effectively reducing the communication failure risk caused by electromagnetic interference, and ensuring the communication security and stable operation of the power network in a complex electromagnetic environment. Brief Description of the Drawings

[0017] Figure 1 is a schematic flowchart of the power network active defense method in an open and integrated network environment provided by an embodiment of the present invention; Figure 2 is a block diagram of the power network active defense system in an open and integrated network environment provided by an embodiment of the present invention. Detailed Embodiments

[0018] The following specifically illustrates the embodiments of the present invention in conjunction with the drawings. The given embodiments are only for illustrative purposes and should not be construed as limiting the present invention. The drawings are only for reference and illustration, and do not constitute a limitation on the protection scope of the present invention patent, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.

[0019] Refer to Figure 1 and an embodiment of the present invention provides a power network active defense method in an open and integrated network environment. As Figure 1 shown, the method includes the following steps: S1. In an open and integrated network environment, analyze the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station, and identify the electromagnetic field intensity gradient change region.

[0020] In this embodiment, the step of analyzing the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station and identifying the electromagnetic field intensity gradient change region includes: Divide the electromagnetic field distribution data among several groups of valve towers collected by the electromagnetic field sensor array in the valve hall of a UHV converter station according to different monitoring regions to obtain the electromagnetic field distribution data of each monitoring region. According to the electromagnetic field distribution data between adjacent monitoring regions, use spatial divergence to calculate the electromagnetic field intensity difference between different monitoring regions. Use a sliding time window to dynamically analyze the electromagnetic field intensity difference, and calculate the change rate of the electromagnetic field intensity gradient between regions. Determine the monitoring regions where the change rate of the electromagnetic field intensity gradient between regions exceeds the preset electromagnetic field intensity gradient change threshold range as the electromagnetic field intensity gradient change regions.

[0021] Specifically, in this embodiment, the electromagnetic field sensor array is uniformly arranged in the valve hall of a UHV converter station in a grid pattern, ensuring that the distance between adjacent electromagnetic field sensors is 3 meters. In this embodiment, a three-axis electromagnetic field sensor is installed at each monitoring point. This layout method can comprehensively and evenly cover the space in the valve hall to collect the electromagnetic field distribution data in the X, Y, and Z directions in real time, ensuring the integrity of the collected data. In this embodiment, a three-axis electromagnetic field sensor is used to monitor and collect the electromagnetic field distribution data at its location in real time. The sampling frequency is set to 100 Hz, and the monitoring time span of the electromagnetic field sensor is 24 hours to obtain sufficient data volume for subsequent analysis. After the deployment of the electromagnetic field sensor array is completed, this embodiment starts to collect the electromagnetic field distribution data in real time. The data acquisition system groups and stores the collected electromagnetic field distribution data according to the monitoring regions (such as around each valve tower, in the center of the valve hall, etc.) to form a time series database for subsequent analysis. Then, this embodiment extracts the electromagnetic field distribution data of adjacent monitoring regions from the time series database and uses the spatial divergence calculation method to analyze the electromagnetic field distribution data between adjacent monitoring regions. For the data at each time point, by calculating the spatial divergence of the electromagnetic field intensity data of adjacent monitoring regions, the electromagnetic field intensity difference between adjacent regions is obtained to quantify the change of the electromagnetic field intensity between different monitoring regions. For example, when the electromagnetic field intensity in monitoring region A is 300 milliteslas, if the electromagnetic field intensity in adjacent region B is 250 milliteslas, the difference between them is 50 milliteslas.

[0022] Next, in this embodiment, 10 minutes is set as the length of the sliding time window. In the time series database, taking 10 minutes as a time segment, the sliding time window method is used to dynamically analyze the difference in electromagnetic field strength. Within each sliding time window, the average change rate of the difference in electromagnetic field strength between adjacent time points is calculated to obtain the gradient change rate of the electromagnetic field strength between regions within each time window, so as to evaluate the dynamic change of the electromagnetic field strength. For example, if within a sliding time window with a length of 10 minutes, the difference in electromagnetic field strength between adjacent regions increases from 50 milliteslas to 70 milliteslas, the change rate is 20 milliteslas / 10 minutes. Finally, in this embodiment, a threshold range for the gradient change of the electromagnetic field strength is preset according to actual engineering experience and requirements such as safety standards for the change of the electromagnetic field strength, and the monitoring regions with significant gradient changes in the electromagnetic field strength are screened out according to the preset threshold range for the gradient change of the electromagnetic field strength. The calculated gradient change rate of the electromagnetic field strength between regions is compared with the preset threshold range. If there is a monitoring region where the gradient change rate of the electromagnetic field strength exceeds the preset threshold range for the gradient change of the electromagnetic field strength, then this monitoring region is determined as an electromagnetic field gradient change region, so as to effectively identify the electromagnetic field gradient change region by accurately analyzing the spatial intensity gradient of the electromagnetic field.

[0023] S2. Analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling strength between each group of valve towers in the electromagnetic field gradient change region, and generate a topological map of electromagnetic interference relationships.

[0024] In this embodiment, the step of analyzing the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling strength between each group of valve towers in the electromagnetic field gradient change region and generating a topological map of electromagnetic interference relationships includes: Construct a vector electromagnetic field model of the valve tower array based on Maxwell's equations, and calculate the electromagnetic field coupling strength between each group of valve towers in the electromagnetic field gradient change region according to the vector electromagnetic field model; Introduce a shielding strength parameter, and calculate the spatial attenuation coefficient of the electromagnetic field between each group of valve towers according to the electromagnetic field coupling strength; According to the spatial attenuation coefficient, use the exponential attenuation function to fit the attenuation amount of the electromagnetic field strength at different propagation distances, and obtain the electromagnetic field propagation attenuation characteristic curve; Calculate the propagation path loss of the electromagnetic wave according to the electromagnetic field propagation attenuation characteristic curve, and use the propagation path loss of the electromagnetic wave as the weight to search for the shortest propagation path between each path node by using Dijkstra's algorithm; Use a graph neural network to analyze the shortest propagation path and electromagnetic field distribution data, calculate the topological correlation degree between each group of valve towers, and construct a topological map of electromagnetic interference relationships between valve towers through the topological correlation degree.

[0025] Specifically, based on Maxwell's equations, this embodiment establishes a partial differential equation describing the electromagnetic field around the valve tower array by combining the specific structural parameters of the valve tower array in the valve hall of the UHV converter station (such as the size, shape, material properties of the valve towers, and their arrangement with each other, etc.). Then, numerical calculation methods such as the finite element method (FEM) or the finite difference method (FDM) are used to solve the partial differential equation to obtain the vector electromagnetic field distribution of the valve tower array. Based on the vector electromagnetic field distribution, a vector electromagnetic field model of the valve tower array is constructed. This vector electromagnetic field model accurately describes the spatial distribution of the vector characteristics of the electric and magnetic fields around the valve tower array, such as the electric and magnetic fields. On the basis of the vector electromagnetic field model, this embodiment uses electromagnetic field theory to calculate the electromagnetic field coupling strength between groups of valve towers in the electromagnetic field intensity gradient region. For example, this embodiment can calculate the magnitude of the electric or magnetic field generated by the electromagnetic field emitted from a single valve tower at an adjacent valve tower, as well as the phase relationship between these two fields, so as to determine their coupling strength. Then, considering the shielding effect between the valve towers, a shielding strength parameter is defined according to the conductivity and magnetism of the valve tower material. The shielding strength parameter can characterize the degree of attenuation of the electromagnetic field during propagation due to the shielding effect. By combining the electromagnetic field coupling strength and the shielding strength parameter, the spatial attenuation coefficient of the electromagnetic field between groups of valve towers is calculated using electromagnetic field propagation theory. According to the spatial attenuation coefficient, the exponential decay function is used to fit the attenuation amount of the electromagnetic field intensity at different propagation distances to obtain the electromagnetic field propagation attenuation characteristic curve. The form of the exponential decay function is expressed as: In the formula, is the electromagnetic field intensity at a propagation distance of d; is the initial electromagnetic field intensity; is the exponential decay coefficient, and its value is equal to the spatial attenuation coefficient.

[0026] Then, this embodiment calculates the electromagnetic wave propagation path loss from one valve tower to another according to the electromagnetic field propagation attenuation characteristic curve. This embodiment can determine the electromagnetic wave propagation path loss by integrating the attenuation situation within a specific distance range of the electromagnetic field propagation attenuation characteristic curve or directly using the fitting function to calculate the attenuation value at the corresponding distance. Then, taking the valve tower array as the nodes in the graph structure, the propagation path as the edges, and the electromagnetic wave propagation path loss as the weight of the edges, the Dijkstra algorithm is used to search for the shortest propagation path from the source node to the target node on the graph structure.

[0027] Finally, preprocess the shortest propagation path and electromagnetic field distribution data that have been searched. For the shortest propagation path data, in this embodiment, it is converted into a graph structure representation, where nodes represent valve towers, edges represent the shortest propagation paths between valve towers, and the weights of the edges are path cost values determined according to electromagnetic wave propagation path loss, etc.; the electromagnetic field distribution data includes numerical information such as the electromagnetic field strength and coupling strength at each valve tower position. After integrating these data in this embodiment, they are used as the input of the graph neural network. In this embodiment, a graph neural network model such as a graph convolutional network (GCN) containing an attention mechanism is used to train the preprocessed data to learn the topological correlation degree between valve towers. The topological correlation degree reflects the tightness between valve towers in terms of electromagnetic field propagation and distribution. The larger the value, the stronger the correlation. The specific process is as follows: In the graph neural network, in this embodiment, first extract the features of the nodes (valve towers). The input node features include information such as the electromagnetic field strength and coupling strength in the electromagnetic field distribution data. The graph neural network generates new node embedding representations by aggregating the feature information of adjacent nodes layer by layer through convolutional operations on these features. This process can help the network learn the correlation between the complex topological relationships between nodes and the electromagnetic field distribution characteristics. For example, the first layer of GCN can calculate the initial embedding vector of each node and update the node representation by aggregating the features of its neighbor nodes and its own features. During the calculation process of the graph neural network, the weights of the edges can be dynamically updated according to the feature embeddings of the nodes. An attention coefficient is calculated for each edge through the attention mechanism, and this coefficient represents the influence degree of the source node on the target node features. This way of dynamically updating the edge weights can better capture the correlation between nodes and the characteristics of the topological structure, so as to calculate a more accurate topological correlation degree. After calculating through multiple layers of the graph neural network, the final embedding representation of each node is obtained. Based on these embedding vectors, calculate the topological correlation degree between each group of valve towers (i.e., each pair of nodes). Specifically, in this embodiment, similarity measurement methods such as cosine similarity can be used to measure the similarity degree between different node embedding vectors, and this is used as a quantization index of the topological correlation degree. In this embodiment, an electromagnetic interference relationship topological graph between valve towers is constructed according to the calculated topological correlation degree. In the topological graph, each node represents a valve tower, the edge represents the electromagnetic interference relationship between valve towers, and the weight of the edge is the value of the corresponding topological correlation degree. It should be noted that in this embodiment, the value of the topological correlation degree needs to be normalized to the interval [0, 1] as the weight of the edge to accurately reflect the strength of the electromagnetic interference relationship, so as to facilitate in-depth analysis of the electromagnetic environment in the valve hall of the UHV converter station.

[0028] S3. Determine the migration direction of the electromagnetic field strength gradient region, and collect the port link quality data between communication devices near the electromagnetic field strength gradient region according to the migration direction of the electromagnetic field strength gradient region.

[0029] In this embodiment, the step of determining the migration direction of the electromagnetic field intensity gradient region includes: Based on the electromagnetic field distribution data, calculate the average field strength of the electromagnetic field intensity gradient region within each time window by using a sliding time window, and obtain a sequence of average field strengths of the electromagnetic field intensity gradient region; Calculate the standard deviation of the field strength within each time window according to the sequence of average field strengths of the electromagnetic field intensity gradient region, and obtain a standard deviation time series; Perform wavelet analysis on the standard deviation time series to extract the electromagnetic field intensity change feature vector; Based on the electromagnetic field intensity change feature vector, use a time series analysis method to model and predict the electromagnetic field intensity change trend, and generate an electromagnetic field intensity change trend matrix; Use a Kalman filter to perform real-time estimation on the electromagnetic field intensity change trend matrix to determine the migration direction of the electromagnetic field intensity gradient region.

[0030] Specifically, in this embodiment, the length of the sliding time window is set based on the electromagnetic field distribution data. The length of the sliding time window can be adjusted according to actual needs, such as set to 1 second, 5 seconds, 10 seconds, etc. Within each time window, the electromagnetic field intensity data of all monitoring points in the electromagnetic field intensity gradient change area are averaged to obtain the average field intensity of the electromagnetic field intensity gradient change area within this time window. The average field intensities corresponding to all time windows are arranged in chronological order to form a sequence of average field intensities in the electromagnetic field intensity gradient change area. For each time window in the sequence of average field intensities in the electromagnetic field intensity gradient change area, the standard deviation of the electromagnetic field intensity data of all monitoring points within this time window is calculated to measure the fluctuation degree of the electromagnetic field intensity within the gradient change area in this time window. The standard deviations of each time window are arranged in chronological order to form a standard deviation time series. Then, wavelet transform is performed on the standard deviation time series. Wavelet transform can decompose the time series into components of different frequencies, thereby extracting characteristic information on different time scales. In this embodiment, by analyzing the coefficients of each layer after wavelet decomposition, characteristic vectors of the electromagnetic field intensity change are extracted. For example, Daubechies wavelet can be used to decompose the standard deviation time series into 3 layers, and the low-frequency and high-frequency components of each layer are extracted as the characteristic vectors of the electromagnetic field intensity change. These characteristic vectors can include information such as the amplitude and phase of different frequency components. Then, in this embodiment, a time series analysis method can be selected according to the time series characteristics of the characteristic vectors of the electromagnetic field intensity change, and the historical characteristic vectors of the electromagnetic field intensity change are used as inputs to train a time series analysis model. The trained time series analysis model is used to predict the electromagnetic field intensity change trend in the next period of time, and the prediction results are organized into an electromagnetic field intensity change trend matrix. Each row of the matrix represents the predicted value at a time point. At the same time, the initial parameters of the Kalman filter are set, such as setting the state vector as the electromagnetic field intensity value, the state transition matrix as the identity matrix, etc. The predicted values of the electromagnetic field intensity change trend matrix are used as observation values and input into the Kalman filter for real-time estimation. The Kalman filter is recursively updated according to the prediction model and the observation model to estimate the state of the electromagnetic field intensity gradient change area in real time. According to the estimation result of the Kalman filter, the migration direction of the electromagnetic field intensity gradient change area is judged. For example, if it is estimated that the position of the electromagnetic field intensity gradient change area moves in a certain direction over time, then this direction is the migration direction. To collect the port link quality data between communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area, the specific steps include: Collect the port data packet transmission rate between communication devices near the electromagnetic field intensity gradient change area based on the migration direction of the electromagnetic field intensity gradient change area; Calculate the port link bandwidth utilization rate according to the port data packet transmission rate and the port physical bandwidth; Based on the port link bandwidth utilization rate and the pre-acquired link quality metrics, obtain the port link quality data between communication devices near the electromagnetic field intensity gradient region.

[0031] Specifically, in this embodiment, according to the determined migration direction of the electromagnetic field intensity gradient region, identify the communication devices located near the migration direction and affected by the electromagnetic field, and use the management interface of the communication devices to collect the port data packet transmission rate data between these communication devices. Divide the port data packet transmission rate by the physical bandwidth to obtain the port link bandwidth utilization rate. Finally, set link quality metrics such as the bandwidth utilization rate threshold according to the requirements of the communication system, and combine the port link bandwidth utilization rate and other link quality metrics into the port link quality data between communication devices near the electromagnetic field intensity gradient region, so as to obtain a more comprehensive and accurate port link quality evaluation result by comprehensively considering multiple factors.

[0032] S4. Obtain the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology map.

[0033] In this embodiment, the step of obtaining the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology map includes: Use the graph mining method to analyze the port link quality data and the electromagnetic interference relationship topology map, and extract the time-varying feature sequences of the port link quality data and the electromagnetic interference relationship topology map at different time points; Adopt the autoregressive integrated moving average (ARIMA) method to perform modeling analysis on the time-varying feature sequences to obtain the topology evolution trend vector; Based on the topology evolution trend vector, use the dynamic mode decomposition algorithm to identify the co-evolution relationship between the port link quality and the electromagnetic field topology structure.

[0034] Specifically, in this embodiment, map the port link quality data to the node attributes of the graph. Each node represents a port or a device, and the attribute value is the link quality metric; at the same time, convert the electromagnetic interference relationship topology map into a graph form, where the nodes represent devices and the edges represent the electromagnetic interference relationships between devices. Use the graph mining method to perform snapshot sampling on the graph at different time points to obtain a series of time-related graph snapshots. Extract features from each snapshot, such as the average link quality of the nodes and the average interference intensity of the edges, and arrange these features in chronological order to form the time-varying feature sequences of the port link quality data and the electromagnetic interference relationship topology map. These time-varying feature sequences record the dynamic changes of the port link quality and the electromagnetic field topology structure over time.

[0035] Then, for the extracted time-varying feature sequence, in this embodiment, the differential autoregressive moving average method is selected for modeling and analysis. In this embodiment, the stationarity of the time-varying feature sequence is tested. If the sequence is non-stationary, a differencing operation is performed until a stationary sequence is obtained. Then, according to the autocorrelation function (ACF) and the partial autocorrelation function (PACF), the orders (p, d, q) of the ARIMA model are determined, where p is the autoregressive order, d is the differencing order, and q is the moving average order. The historical data is used to fit the ARIMA model to obtain the parameter estimation values of the model. In this embodiment, the trained ARIMA model is used to predict the time-varying feature sequence in a future period. The prediction results reflect the evolution trends of the port link quality and the electromagnetic field topology structure. These prediction results are arranged in chronological order to form a topology evolution trend vector. This vector represents the variation trend of the relationship between the port link quality and the electromagnetic interference over time and contains a quantitative description of the future topology structure changes.

[0036] Finally, in this embodiment, the topology evolution trend vectors are arranged in chronological order to construct the input matrix for dynamic mode decomposition. Each column of the input matrix represents the topology evolution trend vector at a time point. In this embodiment, by performing dynamic mode decomposition (DMD) on the topology evolution trend vectors, a series of dynamic mode matrices and corresponding modal matrices are obtained. These dynamic mode matrices reflect the different characteristics and behaviors of the port link quality and the electromagnetic field topology structure during the evolution process. By analyzing the eigenvalues and modal vectors of these dynamic modes, the co-evolution relationship between the port link quality and the electromagnetic field topology structure can be identified. For example, if the eigenvalue of a certain dynamic mode indicates that it has a stable growth trend, and the corresponding modal vector shows that some characteristics of the port link quality index and the electromagnetic field topology structure (such as the interference intensity between specific nodes and the bandwidth utilization rate of the corresponding link) change synchronously, then it can be determined that there is a co-evolution relationship between the port link quality and the electromagnetic field topology structure under this dynamic mode. By comprehensively analyzing all dynamic modes, the co-evolution situation between the two can be fully understood, providing a basis for further optimizing the electromagnetic environment and communication system performance of the UHV converter station.

[0037] S5. Analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security based on the co-evolution relationship between the port link quality and the electromagnetic field topology structure to obtain the communication security risk prediction result.

[0038] In this embodiment, the step of analyzing the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security based on the co-evolution relationship between the port link quality and the electromagnetic field topology structure to obtain the communication security risk prediction result includes: Construct a co-evolution state transition matrix based on the co-evolution relationship between the port link quality and the electromagnetic field topology, and use a long short-term memory network to learn and predict the co-evolution state transition matrix to obtain the bandwidth limitation probability and the packet loss rate; Use the Markov method to predict the states of the bandwidth limitation probability and the packet loss rate to obtain the communication link interruption probability distribution; According to the characteristic parameters of the electromagnetic field intensity gradient change region, use the spatio-temporal correlation analysis method to analyze the dynamic migration of the electromagnetic field intensity gradient change region and construct a spatial position determination matrix; Calculate the overlapping degree between the communication link and the migration path of the gradient change region according to the communication link interruption probability distribution and the spatial position determination matrix; If the overlapping degree exceeds a preset overlapping threshold, calculate the communication link interference probability according to the overlapping degree, and generate a communication security risk prediction result according to the communication link interference probability.

[0039] Specifically, based on the extracted co-evolution relationship between the port link quality and the electromagnetic field topology, this embodiment determines the state space of the system. Each state in the state space can be represented as a combined state, including the key indicators of the port link quality (such as bandwidth utilization rate, packet loss rate, etc.) and the characteristic parameters of the electromagnetic field topology (such as the interference intensity between nodes, topological connection status, etc.). By analyzing historical data, this embodiment statistically calculates the transition probabilities between different states and constructs a co-evolution state transition matrix. The elements of the co-evolution state transition matrix represent the probability of transitioning from one state to another. This embodiment uses the constructed co-evolution state transition matrix as the input data of a long short-term memory network (LSTM) and performs preprocessing operations such as normalization before inputting it into the long short-term memory network. This embodiment uses the trained LSTM model to learn and predict the co-evolution state transition matrix, and predicts the bandwidth limitation probability and the packet loss rate for a future period of time. The bandwidth limitation probability reflects the possible situation of insufficient bandwidth in the communication link in the future, and the packet loss rate reflects the proportion of data packets that may be lost during future transmission.

[0040] Based on the bandwidth - limited probability and packet loss rate predicted by the LSTM network, this embodiment uses the Markov method for state prediction. Specifically, this embodiment takes the bandwidth - limited probability and packet loss rate as the state variables of the system, constructs a Markov chain model, calculates the transition probabilities between various states, and obtains the state transition probability matrix. According to the current bandwidth - limited probability, packet loss rate state, and the state transition probability matrix, it predicts the bandwidth - limited probability and packet loss rate states at each future time point. According to the predicted bandwidth - limited probability state and packet loss rate state, combined with the condition of communication link interruption (such as the link is interrupted when the bandwidth is lower than a certain threshold or the packet loss rate is higher than a certain threshold), it calculates the communication link interruption probability distribution. If it is predicted that the bandwidth - limited probability is high and the packet loss rate also exceeds a certain threshold within a certain time period, then the probability of communication link interruption will increase accordingly. By analyzing the bandwidth - limited probability and packet loss rate at different time points, the probability distribution of communication link interruption within the entire prediction time period is obtained.

[0041] Then, according to the characteristic parameters of the electromagnetic field intensity gradient region (such as intensity gradient, spatial distribution range, migration rate, etc.), this embodiment uses the spatio - temporal correlation analysis method to analyze the dynamic migration law of the electromagnetic field intensity gradient region. The spatio - temporal correlation analysis method can consider the variation law of the electromagnetic field intensity gradient region in space and time, as well as the spatial position relationship with the communication link. Then, based on the analysis results, a spatial position determination matrix is constructed. The spatial position determination matrix is used to judge the possible positions of the future electromagnetic field intensity gradient region. According to the communication network topology structure, the path of the communication link is planned, and the communication link path is compared with the spatial position determination matrix to calculate the overlap degree between the communication link and the migration path of the electromagnetic field intensity gradient region. This embodiment can multiply the communication link interruption probability distribution element - by - element with the spatial position determination matrix, and then perform an integral operation on the multiplication result to obtain an index that comprehensively reflects the degree of interference of the communication link during the migration process in the gradient region, that is, the overlap degree. The higher the overlap degree, the greater the possibility that the communication link is affected by electromagnetic interference during the migration process in the gradient region. If the overlap degree exceeds the preset overlap threshold, it is considered that the communication link may be affected by the electromagnetic field intensity gradient region. Furthermore, by normalizing the overlap degree or using a probability model, the probability of the communication link being affected is calculated, and the communication security risk prediction result is generated according to the affected probability, including risk level, influence range, etc.

[0042] In this embodiment, the step of analyzing the dynamic migration of the electromagnetic field intensity gradient region using the spatio - temporal correlation analysis method according to the characteristic parameters of the electromagnetic field intensity gradient region and constructing a spatial position determination matrix includes: According to the characteristic parameters of the electromagnetic field intensity gradient region, use Gaussian process regression to analyze the spatial distribution state of the electromagnetic field intensity gradient region at different time points, and obtain the dynamic migration trajectory matrix of the electromagnetic field intensity gradient region; the characteristic parameters of the electromagnetic field intensity gradient region include the intensity gradient, spatial distribution range and migration rate of the electromagnetic field intensity gradient region; Based on the dynamic migration trajectory matrix, determine the time point sequence when the magnetic field intensity gradient region changes and its position information at different time points; Use the spatio-temporal correlation analysis method to analyze the time point sequence when the electromagnetic field intensity gradient region changes and its position information at different time points, and obtain the device point density between communication devices near the electromagnetic field intensity gradient region and the link connection degree between communication devices; According to the device point density and the link connection degree, use a weighted algorithm to construct a spatial position determination matrix.

[0043] Specifically, in this embodiment, according to the characteristic parameters of the electromagnetic field intensity gradient region (such as intensity gradient, spatial distribution range, migration rate, etc.), use the Gaussian Process Regression (GPR) method to analyze the spatial distribution state of these characteristic parameters at different time points. GPR is a non-parametric Bayesian regression method that can handle uncertainties in high-dimensional spaces and can be used to analyze the spatio-temporal changes of electromagnetic fields. Specifically, in this embodiment, training data is used to fit the GPR model to obtain the spatial distribution prediction function of the electromagnetic field intensity gradient region. At each time point, this embodiment uses the prediction function to calculate the spatial distribution of the electromagnetic field intensity gradient region, constructs a dynamic migration trajectory matrix, and then traverses the dynamic migration trajectory matrix. By comparing the spatial distribution states of the electromagnetic field intensity gradient region at different time points, determine the time point sequence when the magnetic field intensity gradient region changes. For example, when there are obvious changes in the position, shape or size of the electromagnetic field intensity gradient region between adjacent time points, record this time point as the changed time point, and calculate the center point coordinates and boundary range of the gradient region at each changed time point as the position information.

[0044] In this embodiment, the time point sequence of the change in the electromagnetic field intensity gradient region and its position information at different time points are used as inputs. The spatio-temporal correlation analysis method is used to analyze the communication devices near the electromagnetic field intensity gradient region. The spatio-temporal correlation analysis method can comprehensively consider time and space factors, and dig out the correlation relationship between the electromagnetic field intensity gradient region and the communication devices. At the same time, during the analysis process, the number of communication devices per unit area is calculated near the gradient region to obtain the device point density. At the same time, the number of communication links between the communication devices in the gradient region and the quality of the links (such as bandwidth, delay, etc.) are calculated to obtain the link connectivity. The link connectivity reflects the degree of connection tightness between the communication devices. In this embodiment, corresponding weights are set according to the importance of the device point density and the link connectivity. According to the calculated device point density and link connectivity, a suitable weighted algorithm is selected to construct a spatial position determination matrix. The elements in the matrix can represent the degree to which the communication devices are affected by the electromagnetic field intensity gradient region in terms of spatial position. By comprehensively considering the device point density and the link connectivity, a quantization matrix that can reflect the relationship between the communication devices and the electromagnetic field intensity gradient region in space is obtained. In order to unify the dimension and compare the differences between different communication devices, this embodiment can perform a normalization process on the determination matrix.

[0045] S6. Based on the communication security risk prediction result, generate an optimal power network protection strategy through a reinforcement learning algorithm.

[0046] In this embodiment, the steps of generating an optimal power network protection strategy through a reinforcement learning algorithm based on the communication security risk prediction result include: Combine the power network operation state and the communication security risk prediction result into a multi-dimensional state vector, and use the multi-dimensional state vector as the state space in the reinforcement learning algorithm; Define the protection strategy as the action space according to the protection requirements of the power network, and encode each action in the action space to form a discrete action set; Initialize the Q-value table of the reinforcement learning algorithm. In the power network simulation environment under the open fusion network environment, control the intelligent agent to select and execute actions from the action space according to the current state, and iteratively update the Q-value table according to the rewards obtained from the executed actions. Repeat the above process until the Q-value converges; According to the converged Q-value table, select the action with the maximum Q-value as the optimal protection strategy in the current state.

[0047] Specifically, this embodiment collects the operation status of the power network. The operation status of the power network may include information such as the operation parameters of power equipment (such as voltage, current, power, etc.), network topology structure, power flow distribution, etc. At the same time, the collected data is preprocessed, such as normalization, standardization, etc., to ensure that data with different dimensions can be compared on the same scale. The processed operation status data of the power network and the communication security risk prediction results are combined into a multi-dimensional state vector. This multi-dimensional state vector can comprehensively describe the operation status of the power network at a specific moment and the communication security risk situation faced. The multi-dimensional state vector is used as the state space in the reinforcement learning algorithm. Each state point in the state space corresponds to a specific operation status and communication security risk situation of the power network at a certain moment, providing a decision-making basis for the agent. A series of protection strategies are defined according to the protection requirements of the power network. The protection strategies may include adjusting voltage, current limit, enabling a firewall, deploying an intrusion detection system, updating security patches, etc. Each protection strategy in the action space is encoded to form a discrete action set. The encoding can adopt digital encoding, or other suitable encoding methods such as binary encoding and symbol encoding. For example, when using digital encoding, this embodiment can use the number 1 to represent the voltage adjustment strategy, the number 2 to represent the firewall enabling strategy, and so on. The formed discrete action set is used as the action space in the reinforcement learning algorithm, that is, the set of all possible actions that the agent can execute when making a decision. Then this embodiment creates a Q-value table. The Q-value table is used to record the expected rewards obtained by the agent for executing each action in each state. Initially, all values in the Q-value table can be set to 0 or a certain random value, indicating that in the initial state, the value evaluation for each state-action pair is the same.

[0048] In this embodiment, a power network simulation environment is built in an open and integrated network environment. This simulation environment can simulate the actual operation of the power network and update and feedback the state of the power network according to the actions executed by the agent. This embodiment simulates the real operation state and communication security risks of the power network through the built power network simulation environment. In the simulation environment, the control agent selects an action to execute from the action space according to the current multi-dimensional state vector. When selecting an action, this embodiment can use strategies such as the ε-greedy strategy to select the action of the agent. According to the change in the operation state of the power network and the degree of reduction in communication security risks after the agent executes the action, the reward value obtained by the agent is calculated. The reward function is related to objectives such as the operation state of the power network and the degree of reduction in communication security risks. The reward can be a positive number (indicating a reduction in security risks) or a negative number (indicating an increase in security risks or a decrease in the performance of the power network). For example, when the interference probability of the communication link decreases and the operation efficiency of the power equipment increases after executing a certain action, a higher reward value is given; conversely, if the interruption probability of the communication link increases and the power equipment fails, etc., a lower reward value is given. Finally, according to the obtained reward value, the current state, and the action, reinforcement learning algorithms such as the Deep Q-Network (DQN) are used to update the Q-value table. The above process is repeated until the Q-value table converges, that is, the change in the Q-value tends to be stable. The judgment condition for the convergence of the Q-value can be set as that in continuous several iterations, the change in the Q-value in the Q-value table is less than a preset threshold. After the Q-value table converges, in the row corresponding to the current multi-dimensional state vector, the action with the largest Q-value is selected as the optimal protection strategy in the current state, which can minimize the communication security risks and ensure the stable operation of the power network. This embodiment applies the selected optimal protection strategy to the actual power network, and makes corresponding adjustments and optimizations to the operation state of the power network and the communication system to reduce the communication security risks and effectively improve the communication security and reliability of the power network in complex situations such as dynamic migration in the electromagnetic field intensity gradient change area.

[0049] An embodiment of the present invention provides a method for active defense of a power network in an open and integrated network environment. The method analyzes the electromagnetic field space intensity gradient based on the electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station, and identifies the electromagnetic field intensity gradient change area; analyzes the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area, and generates an electromagnetic interference relationship topology diagram; collects the port link quality data among the communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area; obtains the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology diagram; analyzes the potential impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtains the communication security risk prediction result; based on the communication security risk prediction result, generates an optimal power network protection strategy through a reinforcement learning algorithm. Compared with the prior art, this method accurately analyzes the impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security by monitoring the electromagnetic field distribution in the valve hall of the UHV converter station, and combines the electromagnetic interference relationship topology diagram and the co-evolution relationship to achieve the active defense of the power network in an open and integrated environment, effectively reducing the communication failure risk caused by electromagnetic interference, and ensuring the communication security and stable operation of the power network in a complex electromagnetic environment.

[0050] It should be noted that the magnitudes of the sequence numbers of the above processes do not mean the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0051] In one embodiment, as Figure 2 shown, an embodiment of the present invention provides an active defense system for a power network in an open and integrated network environment, and the system includes: A gradient change identification module 101, configured to analyze the electromagnetic field space intensity gradient based on the electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station in an open and integrated network environment, and identify the electromagnetic field intensity gradient change area; An electromagnetic analysis module 102, configured to analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area, and generate an electromagnetic interference relationship topology diagram; A migration analysis module 103, configured to determine the migration direction of the electromagnetic field intensity gradient change area, and collect the port link quality data among the communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area; A co-evolution module 104, configured to obtain the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology diagram; A risk prediction module 105 is configured to analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security based on the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtain a communication security risk prediction result. A policy generation module 106 is configured to generate an optimal power network protection policy through a reinforcement learning algorithm based on the communication security risk prediction result.

[0052] Specific limitations on an active defense system for a power network in an open and integrated network environment can refer to the above limitations on an active defense method for a power network in an open and integrated network environment, which will not be elaborated here. Those of ordinary skill in the art can realize that, combining the various modules and steps described in the embodiments disclosed in this application, they can be implemented in hardware, software, or a combination of both. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0053] The embodiment of the present invention provides an active defense system for a power network in an open and integrated network environment. The gradient change recognition module of the system analyzes the electromagnetic field space intensity gradient based on the monitored electromagnetic field distribution data between several groups of valve towers in the valve hall of a UHV converter station, and identifies the electromagnetic field intensity gradient change region; the electromagnetic analysis module analyzes the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity between the valve towers in the electromagnetic field intensity gradient change region, and generates an electromagnetic interference relationship topology graph; the migration analysis module collects the port link quality data between communication devices near the electromagnetic field intensity gradient change region according to the migration direction of the electromagnetic field intensity gradient change region; the co-evolution module obtains the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology graph; the risk prediction module analyzes the potential impact of the dynamic migration of the electromagnetic field intensity gradient change region on communication security based on the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtains a communication security risk prediction result; the policy generation module generates an optimal power network protection policy through a reinforcement learning algorithm based on the communication security risk prediction result. Compared with the prior art, the system accurately analyzes the impact of the dynamic migration of the electromagnetic field intensity gradient change region on communication security by monitoring the electromagnetic field distribution in the valve hall of the UHV converter station, and combines the electromagnetic interference relationship topology graph and the co-evolution relationship to achieve the active defense of the power network in an open and integrated environment, effectively reducing the communication failure risk caused by electromagnetic interference, and ensuring the communication security and stable operation of the power network in a complex electromagnetic environment.

[0054] The above-described embodiments merely represent several preferred embodiments of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and substitutions can be made, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the protection scope of the claims.

Claims

1. An active defense method for power networks in an open and integrated network environment, characterized in that Including the following steps: In an open and integrated network environment, analyze the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station, and identify the electromagnetic field intensity gradient change area; Analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area, and generate an electromagnetic interference relationship topology diagram; Determine the migration direction of the electromagnetic field intensity gradient change area, and collect the port link quality data among the communication devices near the electromagnetic field intensity gradient change area according to the migration direction of the electromagnetic field intensity gradient change area; According to the port link quality data and the electromagnetic interference relationship topology diagram, obtain the co-evolution relationship between the port link quality and the electromagnetic field topology structure; Analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient change area on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtain the communication security risk prediction result; Based on the communication security risk prediction result, generate an optimal power network protection strategy through a reinforcement learning algorithm.

2. The active defense method for a power grid in an open and converged network environment according to claim 1, characterized in that, The step of analyzing the electromagnetic field spatial intensity gradient based on the monitored electromagnetic field distribution data among several groups of valve towers in the valve hall of a UHV converter station and identifying the electromagnetic field intensity gradient change area includes: Divide the electromagnetic field distribution data among several groups of valve towers collected by the electromagnetic field sensor array in the valve hall of the UHV converter station according to different monitoring areas to obtain the electromagnetic field distribution data of each monitoring area; According to the electromagnetic field distribution data between adjacent monitoring areas, calculate the electromagnetic field intensity difference between different monitoring areas by using spatial divergence; Dynamically analyze the electromagnetic field intensity difference by using a sliding time window, and calculate the change rate of the electromagnetic field intensity gradient between regions; Determine the monitoring area where the change rate of the electromagnetic field intensity gradient between regions exceeds the preset electromagnetic field intensity gradient change threshold range as the electromagnetic field intensity gradient change area.

3. The active defense method for power network in an open and integrated network environment according to claim 1, characterized in that, The step of analyzing the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area and generating an electromagnetic interference relationship topology diagram includes: Construct a vector electromagnetic field model of the valve tower array based on Maxwell's equations, and calculate the electromagnetic field coupling intensity among the valve towers in the electromagnetic field intensity gradient change area according to the vector electromagnetic field model; Introduce a shielding intensity parameter, and calculate the spatial attenuation coefficient of the electromagnetic field between each group of valve towers according to the electromagnetic field coupling intensity; According to the spatial attenuation coefficient, use an exponential attenuation function to fit the attenuation amount of the electromagnetic field intensity at different propagation distances, and obtain the electromagnetic field propagation attenuation characteristic curve; Calculate the electromagnetic wave propagation path loss according to the electromagnetic field propagation attenuation characteristic curve, and use the electromagnetic wave propagation path loss as the weight to search for the shortest propagation path between each path node by using Dijkstra's algorithm; Use a graph neural network to analyze the shortest propagation path and the electromagnetic field distribution data, calculate the topological correlation degree between each group of valve towers, and construct an electromagnetic interference relationship topology diagram between the valve towers through the topological correlation degree.

4. The active defense method for power grid under an open and integrated network environment according to claim 1, wherein The step of determining the migration direction of the electromagnetic field intensity gradient change area includes: Based on the electromagnetic field distribution data, use a sliding time window to calculate the average field strength of the electromagnetic field intensity gradient region within each time window, and obtain a sequence of the average field strength of the electromagnetic field intensity gradient region; Calculate the standard deviation of the field strength within each time window according to the sequence of the average field strength of the electromagnetic field intensity gradient region, and obtain a standard deviation time series; Perform wavelet analysis on the standard deviation time series to extract the electromagnetic field intensity change feature vector; Based on the electromagnetic field intensity change feature vector, use a time series analysis method to model and predict the electromagnetic field intensity change trend, and generate an electromagnetic field intensity change trend matrix; Use a Kalman filter to perform real-time estimation on the electromagnetic field intensity change trend matrix to determine the migration direction of the electromagnetic field intensity gradient region.

5. The active defense method for power network in an open and converged network environment according to claim 1, wherein The step of collecting the port link quality data between communication devices near the electromagnetic field intensity gradient region according to the migration direction of the electromagnetic field intensity gradient region includes: Collect the port data packet transmission rate between communication devices near the electromagnetic field intensity gradient region based on the migration direction of the electromagnetic field intensity gradient region; Calculate the port link bandwidth utilization rate according to the port data packet transmission rate and the port physical bandwidth; Obtain the port link quality data between communication devices near the electromagnetic field intensity gradient region according to the port link bandwidth utilization rate and the pre-acquired link quality indicators.

6. The active power grid defense method in an open and integrated network environment according to claim 1, characterized in that, The step of obtaining the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology map includes: Use a graph mining method to analyze the port link quality data and the electromagnetic interference relationship topology map, and extract the time-varying feature sequences of the port link quality data and the electromagnetic interference relationship topology map at different time points; Adopt the autoregressive integrated moving average method to perform modeling analysis on the time-varying feature sequences to obtain a topology evolution trend vector; Based on the topology evolution trend vector, use the dynamic mode decomposition algorithm to identify the co-evolution relationship between the port link quality and the electromagnetic field topology structure.

7. The active defense method for power grid under an open and converged network environment according to claim 1, characterized in that The step of analyzing the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure and obtaining the communication security risk prediction result includes: Construct a co-evolution state transition matrix according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and use a long short-term memory network to perform learning and prediction on the co-evolution state transition matrix to obtain the bandwidth limitation probability and the packet loss rate; Use the Markov method to perform state prediction on the bandwidth limitation probability and the packet loss rate to obtain the communication link interruption probability distribution; According to the characteristic parameters of the electromagnetic field intensity gradient region, use a spatio-temporal correlation analysis method to analyze the dynamic migration of the electromagnetic field intensity gradient region, and construct a spatial position determination matrix; Calculate the overlapping degree between the communication link and the migration path of the gradient region according to the communication link interruption probability distribution and the spatial position determination matrix; If the overlapping degree exceeds a preset overlapping threshold, then calculate the communication link interference probability according to the overlapping degree, and generate a communication security risk prediction result according to the communication link interference probability.

8. The active defense method for power network in an open and integrated network environment according to claim 7, characterized in that, The steps of analyzing the dynamic migration of the electromagnetic field intensity gradient region according to the characteristic parameters of the electromagnetic field intensity gradient region and constructing a spatial position determination matrix by using a spatio-temporal correlation analysis method include: According to the characteristic parameters of the electromagnetic field intensity gradient region, using Gaussian process regression to analyze the spatial distribution state of the electromagnetic field intensity gradient region at different time points, and obtaining the dynamic migration trajectory matrix of the electromagnetic field intensity gradient region; the characteristic parameters of the electromagnetic field intensity gradient region include the intensity gradient, spatial distribution range, and migration rate of the electromagnetic field intensity gradient region; Based on the dynamic migration trajectory matrix, determine the time point sequence at which the magnetic field intensity gradient region changes and its position information at different time points; Using the spatio-temporal correlation analysis method to analyze the time point sequence at which the electromagnetic field intensity gradient region changes and its position information at different time points, and obtaining the device point density between communication devices near the electromagnetic field intensity gradient region and the link connection degree between communication devices; According to the device point density and the link connection degree, use a weighted algorithm to construct a spatial position determination matrix.

9. The active power grid defense method in an open and integrated network environment according to claim 1, characterized in that, The steps of generating an optimal power network protection strategy through a reinforcement learning algorithm based on the communication security risk prediction result include: Combining the power network operation state and the communication security risk prediction result into a multi-dimensional state vector, and using the multi-dimensional state vector as the state space in the reinforcement learning algorithm; Defining protection strategies as the action space according to the protection requirements of the power network, and encoding each action in the action space to form a discrete action set; Initializing the Q-value table of the reinforcement learning algorithm. In the power network simulation environment under the open fusion network environment, control the intelligent agent to select and execute actions from the action space according to the current state, and iteratively update the Q-value table according to the rewards obtained from the executed actions. Repeat the above process until the Q-value converges; According to the converged Q-value table, select the action with the maximum Q-value as the optimal protection strategy in the current state.

10. An active power grid defense system in an open and integrated network environment, characterized in that, The system includes: A gradient change recognition module, which is used to analyze the electromagnetic field space intensity gradient according to the electromagnetic field distribution data between several groups of valve towers in the valve hall of the UHV converter station detected in the open fusion network environment, and identify the electromagnetic field intensity gradient region; An electromagnetic analysis module, which is used to analyze the spatial propagation attenuation characteristics of the electromagnetic field according to the electromagnetic field coupling intensity between groups of valve towers in the electromagnetic field intensity gradient region, and generate an electromagnetic interference relationship topology diagram; A migration analysis module, which is used to determine the migration direction of the electromagnetic field intensity gradient region, and collect the port link quality data between communication devices near the electromagnetic field intensity gradient region according to the migration direction of the electromagnetic field intensity gradient region; A co-evolution module, which is used to obtain the co-evolution relationship between the port link quality and the electromagnetic field topology structure according to the port link quality data and the electromagnetic interference relationship topology diagram; A risk prediction module, which is used to analyze the potential impact of the dynamic migration of the electromagnetic field intensity gradient region on communication security according to the co-evolution relationship between the port link quality and the electromagnetic field topology structure, and obtain the communication security risk prediction result; A strategy generation module, configured to generate an optimal power network protection strategy based on the communication security risk prediction result through a reinforcement learning algorithm.

Images