Satellite node reliability evaluation method in F6G low earth orbit satellite optical network

Abstract

The application provides a satellite node reliability evaluation method in a F6G low-orbit satellite optical network, including the following steps: initializing a low-orbit satellite optical network topology to obtain a static topology graph corresponding to each time slot; collecting satellite element state information and environment data to calculate a satellite failure probability; using two-line orbit data of the satellite and orbit debris to calculate a running position and a relative motion state of the satellite, and calculating a collision probability of the satellite and the orbit debris; calculating a time weight and an importance value of each time slot according to the static topology graph, and obtaining a node importance of the satellite through weighted fusion; preprocessing the obtained satellite failure probability, collision probability and node importance; taking the preprocessed data as input of a fuzzy comprehensive evaluation model, and dividing the satellite into satellite nodes with different risks according to a comprehensive evaluation result. The application can comprehensively evaluate the reliability of the satellite, provide targeted protection to ensure the stability of the satellite operation, and reduce the influence risk of the failure on communication.

Description

Technical Field

[0001] This invention relates to the technical field of satellite risk prediction in low-Earth orbit satellite optical networks, and more particularly to a method for evaluating the reliability of satellite nodes in an F6G low-Earth orbit satellite optical network. Background Technology

[0002] With the acceleration of digitalization and the emerging development of the aerospace industry, low-Earth orbit satellite networks, as a supplement to terrestrial networks, provide communication services and seamless global coverage to remote areas, oceans, aviation, and other areas lacking coverage, promising to achieve high-bandwidth, low-latency, and highly reliable network services for terrestrial users. However, with the rapid increase in the number of space debris and spacecraft in the space environment, their relative speeds can reach an average of 10 km / s, increasing the risk of component failures and other malfunctions. Collisions pose a significant threat to the normal flight of spacecraft, leading to a deterioration in satellite communication service quality. Early detection of satellite reliability allows for focused attention on these less reliable satellites, improving service quality. Simultaneously, random failure of satellite nodes is an important research issue, primarily caused by aging satellite components and environmental factors, with the probability of failure increasing over time. In the fault-prone and resource-constrained space environment, a comprehensive and accurate method for assessing satellite node reliability is needed.

[0003] Existing satellite node evaluation methods generally fall into two categories: (1) Reliability assessment methods for individual satellites or components, which are usually based on device lifetime models and historical failure data, using fault tree analysis, Weibull distribution, and other means to estimate failure rates and lifetime indicators. This method can describe the physical failure mechanism of a single satellite or key component, but it easily overlooks the topological position and service carrying role of a single satellite in the entire constellation. Therefore, this method is insufficient in describing the impact on network-level services. (2) Node importance assessment methods based on complex network theory, which are based on topological indicators such as degree centrality, betweenness centrality, and local triangle centrality, or introduce multi-indicator weighting for comprehensive scoring. This method can identify key nodes that are crucial to connectivity and routing from the perspective of network structure. However, this method is based on static topology and cannot reflect the highly dynamic characteristics of low-orbit satellite optical networks. Moreover, it lacks comprehensive consideration of factors such as random failure probability and space debris collision risk. Therefore, this method is still insufficient in assessing the reliability of satellite nodes.

[0004] Chinese patent application CN202311492842.9 discloses a method for risk assessment of satellite impacts from the space environment. The method includes: collecting data on the environmental impacts on the satellite in space, conducting an operational environment interference risk assessment on the environmental impact data, understanding the impact of the environmental data on the satellite, performing trend prevention assessment analysis if a normal instruction is received, and simultaneously conducting early warning operations if a risk instruction is received; collecting sensor status data and transmission data, and conducting operational safety assessment operations on the status data analysis; performing transmission delay risk analysis on the transmission data through information feedback; and performing protection management matching analysis based on the received risk instructions and delay instructions, combining data transmission delay information to perform reasonable environmental protection management matching, thereby reducing the risk of space environment impacts on the satellite. This invention can improve the accuracy of the analysis results, ensuring the safety and stability of satellite operation and reducing the risk of space environment impacts on the satellite. However, this invention mainly addresses the risk status of the satellite's outer space environment without considering specific risk types, and cannot provide high-quality services for satellite network communication. Summary of the Invention

[0005] To address the technical problem of reduced service quality caused by satellite failures in existing low-Earth orbit (LEO) satellite optical networks, this invention proposes a satellite node reliability evaluation method for F6G LEO satellite optical networks. This method utilizes fuzzy comprehensive evaluation to assess satellite node reliability, classifying satellite nodes into high-risk, moderate-risk, and low-risk categories. Nodes with high-risk values ​​receive focused attention during protection efforts, improving the accuracy of the evaluation and the reliability of satellite operation. This invention can comprehensively assess satellite reliability and provide targeted protection to ensure satellite operational stability and reduce the impact of failures on communication.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: a method for evaluating the reliability of satellite nodes in an F6G low-Earth orbit satellite optical network, comprising the following steps:

[0007] Step 1: Initialize the low-Earth orbit satellite optical network topology, divide the orbital period into several time slots to obtain the static topology map corresponding to each time slot; collect satellite component status information and environmental data, and calculate the satellite failure probability;

[0008] Step 2: Calculate the satellite's position and relative motion state using the two lines of orbital data from the satellite and the orbital debris. Based on the spatial positional relationship and relative motion geometry between the satellite and the orbital debris at the rendezvous point, calculate the probability of collision between the satellite and the orbital debris.

[0009] Step 3: Calculate the time weight and importance value of each time slot based on the static topology diagram, and obtain the node importance of the satellite through weighted fusion;

[0010] Step 4: Preprocess the obtained satellite failure probability, collision probability, and node importance;

[0011] Step 5: Use the preprocessed data as input to the fuzzy comprehensive evaluation model to obtain the comprehensive evaluation results. Based on the comprehensive evaluation results, classify the satellites into three categories: high-risk, general-risk, and low-risk satellite nodes.

[0012] Preferably, the method for initializing the low-Earth orbit satellite optical network topology is as follows: satellites are abstracted as nodes, inter-satellite links are abstracted as edges, and the low-Earth orbit satellite optical network topology is abstracted as an undirected time-spreading graph. , where the set of nodes , This indicates the total number of nodes in the satellite optical network. Represents the set of inter-satellite links, and Represents a node and nodes Inter-satellite links; adjacency matrix of nodes , among which, element yes Represents a node and nodes Inter-satellite links exist, otherwise ;

[0013] According to the time spread diagram theory, the satellite's orbital period The topology is divided into a series of snapshots. Where K represents the total number of time slots, and the k-th time slot is represented as... ,in, This represents the start time of the selected time period. This represents the end time of the selected time period; within the orbital period. Inside, in the Each time slot Static topology diagram of the built-in satellite optical network The entire orbital period The network topology diagram is represented as .

[0014] Preferably, the satellite component status information includes operating parameters and historical fault records. The operating parameters include the operating status information of the satellite node's computing resources, storage resources, and link bandwidth. The historical fault records are formed during the long-term operation of the satellite system. The environmental data is data describing the satellite's operating environment and orbital characteristics.

[0015] Missing and outlier values ​​in historical fault records are processed, and normalization methods are used to map working parameters of different dimensions to a comparable numerical range. Then, various working parameters and historical fault data are organized into a satellite component state characteristic sequence according to time order and corresponding fault markers. Satellite faults include early fault period, random fault period and wear and tear fault period.

[0016] Preferably, the method for calculating the satellite failure probability is as follows: inputting the satellite component state feature sequence composed of preprocessed satellite component state information into the failure probability model to obtain the failure probability of the satellite component.

[0017] Preferably, the fault probability model employs a Long Short-Term Memory (LSTM) network with an attention mechanism, comprising an input layer, a CNN layer, an LSTM layer, an attention layer, and an output layer connected in sequence. The input layer receives the satellite component state feature sequence; the CNN layer performs local convolution operations on the input satellite component state feature sequence to extract local features at different times and scales; the LSTM layer models the dependencies of the feature sequence over time; the attention layer includes an attention mechanism and a fully connected layer. The attention mechanism calculates weight coefficients based on the importance of different time steps or features to fault determination, weighting the output features of the LSTM layer; the fully connected layer performs a linear transformation on the weighted features; and the output layer maps the result of the attention mechanism to the corresponding fault probability and outputs it.

[0018] Preferably, the probability density function of the velocity increment of the orbital debris follows a normal distribution;

[0019] The internal coincidence method is used to analyze two lines of orbit data to find the time period of the quasi-reference orbit, so that the error between the reference orbit and the actual orbit is minimized during that time period;

[0020] Using the selected orbit prediction algorithm, the position vector and velocity vector in the inertial coordinate system are calculated at a given time based on the two lines of orbit data of the satellite and the orbital debris. When the spatial distance between the satellite and the orbital debris is less than a pre-set distance threshold at a certain time, it is considered that there is a potential collision risk between the satellite and the orbital debris, and the collision probability is calculated.

[0021] Preferably, the collision probability for:

[0022]

[0023]

[0024]

[0025]

[0026] in, The radial height difference between the satellite and orbital debris when they cross the line of intersection of their common orbital plane; It is a unit vector along the direction of relative position; The distance between the satellite and the intersection line when the orbital debris arrives. The angle between the track surfaces; This represents the relative position vector of the satellite and orbital debris at the moment of their encounter; Located in the plane of relative motion, and relative to the unit vector A vertical tangential unit vector; It is the unit normal vector of the plane of relative motion; The collision detection radius, Indicates the velocity increment of the fragment The decimal logarithm.

[0027] Preferably, the method for calculating the node importance of the satellite is as follows: in the static topology graph of each time slot, the node degree centrality, betweenness centrality, and local triangle centrality are selected to calculate the initial importance of the node; based on the initial importance, proximity centrality, and distance between nodes, an importance contribution matrix is ​​constructed, and the importance value of the node in each time slot is calculated through the importance contribution matrix; the time weight is determined according to the proportion of the time slot duration to the entire orbital period, and the importance values ​​of each time slot are weighted and fused to obtain the comprehensive importance of the node in the entire orbital period, thus obtaining the node importance of the satellite.

[0028] Preferably, the method for calculating the importance value of the computing node in each time slot is as follows:

[0029] In each time slot Static topology graph Taking into account the global importance of a node and the contributions of its neighbors, the nodes are... initial importance Combined with the importance contribution matrix, the i-th node is obtained. In the time slot Importance value for:

[0030]

[0031] The time weight for:

[0032] ;

[0033] For all The satellite's nodal importance is obtained by time-weighted summation of the importance values ​​of each time slot:

[0034]

[0035] in, Represents a node and neighboring nodes The shortest distance between them Representing neighboring nodes Approaching the center, For neighboring nodes The initial importance; For time slots The duration.

[0036] Preferably, the preprocessing in step 4 involves normalizing the data on satellite failure probability, collision probability, and node importance to a unified dimension and direction, mapping them to the same numerical range, and truncating or smoothing outliers; the preprocessed data on satellite failure probability, collision probability, and node importance are then used as a vector input into the fuzzy comprehensive evaluation model.

[0037] The fuzzy comprehensive evaluation model uses satellite failure probability, collision probability, and node importance as evaluation factors to form a factor set; it uses three satellite types—high-risk, moderate-risk, and low-risk—as evaluation sets; based on the preprocessed index values ​​of satellite failure probability, collision probability, and node importance, it constructs membership functions for each evaluation factor corresponding to different risk levels, calculates the membership degree of each satellite at each risk level, and forms a membership degree matrix; it determines the weight vector based on the relative importance of each evaluation factor in satellite reliability assessment, performs weighted operations on the membership degree matrix according to the weights, and obtains the comprehensive membership degree for each risk level; based on the obtained comprehensive membership degree, it uses the maximum membership degree principle to determine the risk level to which the satellite belongs, classifying the satellites into three categories: high-risk, moderate-risk, and low-risk.

[0038] The beneficial effects of this invention are as follows: First, the satellite optical network topology is initialized. Based on the given constellation and orbital parameters, a time-spread graph is constructed with satellites as nodes and inter-satellite optical links as edges. The orbital period is divided into several time slots, and a static topology snapshot corresponding to each time slot is obtained. Simultaneously, satellite component status information and environmental data are collected through onboard sensors or ground monitoring systems, and the risk level of different satellite environments is quantitatively assessed using a space environment risk model. Second, the collected component status information is used as model input to calculate the satellite failure probability. Various component status variables are uniformly mapped to failure probabilities to quantitatively reflect the possibility of random satellite failures. Furthermore, a failure probability model is established and calibrated based on historical failure data. The model is trained and corrected using historical operation and failure records, thereby improving the rationality of the current failure probability estimate. The invention improves the comprehensiveness and accuracy of satellite reliability assessment. Next, it uses two lines of orbital data from the satellite and orbital debris to calculate the satellite's position and relative motion state, obtaining the collision probability between the satellite and debris within a given time window. This probability is then compared with a preset threshold to determine if a collision risk exists. Next, it calculates the time weight within the network topology cycle to obtain the importance data of satellite nodes. Based on the duration of each time slice within a topology cycle, it weights and summarizes the role of each node in the network topology and service transmission, obtaining the importance index of the satellite nodes. This identifies satellite nodes that are more critical to network connectivity and service carrying capacity throughout the cycle. The obtained fault probability, collision probability, and importance index are preprocessed and used as input to a fuzzy comprehensive evaluation model. The fuzzy comprehensive evaluation method is then used to evaluate satellite reliability, classifying satellites into three categories: high-risk, moderate-risk, and low-risk. This invention improves the comprehensiveness and accuracy of satellite reliability assessment, optimizes satellite network communication service management, ensures stable operation, and reduces fault risks. This invention improves the accuracy of reliability assessment in low-Earth orbit satellite optical network fault scenarios, helping to make reasonable and targeted protective measures for satellite network communication service management, ensuring the stability of satellite operation and reducing the impact of faults on communication. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart of the present invention.

[0041] Figure 2 This is a schematic diagram of the satellite component failure model in this invention.

[0042] Figure 3 This is a schematic diagram of the satellite collision probability in this invention.

[0043] Figure 4 This is a schematic diagram of the node reliability time spread graph in this invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] like Figure 1 As shown, a satellite node reliability evaluation method in an F6G low-Earth orbit satellite optical network mainly consists of three parts: satellite failure probability calculation, satellite and debris collision probability calculation, and satellite node importance calculation. The specific implementation steps of this invention are as follows:

[0046] Step 1: Initialize the low-Earth orbit satellite optical network topology, divide the orbital period into several time slots to obtain the static topology map corresponding to each time slot; collect satellite component status information and environmental data, and calculate the satellite failure probability.

[0047] Initialize a low-Earth orbit (LEO) satellite optical network. The topology of the LEO satellite optical network is finite and exhibits a periodic variation pattern. Satellites can be abstracted as nodes, and inter-satellite links can be abstracted as edges. Based on the given constellation and orbital parameters, the LEO satellite optical network topology is abstracted as an undirected time-spreading graph. This is used to represent a satellite network topology diagram, where the set of nodes... , This represents the total number of nodes in the satellite optical network. Represents the set of inter-satellite links, and Indicates satellite node and satellite Inter-satellite links between nodes. The adjacency matrix of a node is defined as follows: In this matrix, the number of rows and columns are equal to the total number of nodes. ,element yes Represents a node and nodes Inter-satellite links exist, otherwise And define and Inter-satellite links The transmission delay and link bandwidth. Satellite nodes operate periodically, dividing the orbital period into several time slots. In each time slot, the network topology is static and remains unchanged. According to the time spread graph theory, the satellite's orbital period... The topology is divided into a series of snapshots. Where K represents the total number of time slots, and one time slot is represented as K = K + K. ,in, This is the start time of the selected time period for this reliability assessment. This represents the end time of the selected time period. (During the orbital period) Within, for each time slot Establish the corresponding static topology snapshot, i.e., the static topology graph. The entire orbital period The network topology diagram is represented as ,in, Indicates the first Each time slot A static topology diagram of the internal satellite optical network. Subsequent calculations of satellite node importance and reliability will use these static topologies. Based on this, the index is calculated separately for each time slot, and the results are combined with the time weight of each time slot over the entire orbital period to reflect the comprehensive role of satellite nodes in the time-varying network.

[0048] Real-time status information and environmental data of each satellite component are collected via data interfaces from onboard sensors or ground monitoring systems. The real-time status information includes operating parameters and historical fault records. Operating parameters include operational status information such as the satellite node's computing resources, storage resources, and link bandwidth. Historical fault records are generated during the long-term operation of the satellite system. Environmental data describes the satellite's operating environment and orbital characteristics. The satellite component status information is used as model input in subsequent steps to estimate the satellite's failure probability; environmental and orbital data are used in subsequent steps to estimate risk indicators such as the probability of collisions between the satellite and debris.

[0049] After collecting historical fault data of components from the satellite system, the raw data is processed for missing and outlier values. A unified normalization method is used to map operating parameters of different dimensions to a comparable numerical range. Then, various operating parameters and historical fault data are organized chronologically with corresponding fault markers to form a satellite component state feature sequence. Representative and fault-sensitive features are extracted to provide standardized input for training subsequent fault probability models. Satellite faults include early fault periods, random fault periods, and attrition fault periods. To more objectively assess the fault probability of satellite nodes, this invention employs a Long Short-Term Memory (LSTM) network with an attention mechanism, such as... Figure 2As shown, the system comprises an input layer, a CNN layer, an LSTM layer, an attention layer, and an output layer connected in sequence. The input layer receives the preprocessed satellite component state feature sequence; the CNN layer performs local convolution operations on the input features to extract local features at different times and scales; the LSTM layer models the dependencies of the feature sequence over time; the attention layer includes an attention mechanism and a fully connected layer. The attention mechanism calculates weight coefficients based on the importance of different features for fault determination at different time steps, and the fully connected layer performs a linear transformation on the weighted features. The attention layer weights the output features of the LSTM layer, highlighting features more critical for fault determination; the output layer maps the results of the attention layer to the corresponding fault probability output. After acquiring the data, a Long Short-Term Memory (LSTM) network is used to time-series the collected satellite component state information. This involves arranging the operating parameters of the same satellite component at multiple consecutive times according to the sampling time order, forming an input sequence reflecting the changes in satellite component state over time. This sequence serves as the model input, capturing long-term dependencies in the time-series data. Combined with a satellite component reliability model, continuous monitoring and updates enable real-time assessment of the satellite network's health status. The reliability model describes the reliability status of a component at the current stage. Based on the division of the component's lifetime into early failure period, random failure period, and wear-out failure period, the network output is mapped to the reliability level of each stage. Through the forward computation of the network, the output value of the corresponding satellite component within each time window is obtained, and this output value is used as the failure probability of the satellite component within that time window, thus obtaining the failure probability of each satellite component.

[0050] Historical fault data is compiled into a dataset and divided into training, validation, and test sets. The training set is used to learn model parameters, the validation set is used to monitor model performance and tune hyperparameters during training, and the test set is used to evaluate the model's generalization ability on unknown data after training. The network structure of a Long Short-Term Memory (LSTM) network incorporating an attention mechanism can automatically assign weights to different fault-related features while capturing long-term dependencies in component state time series, thereby improving the accuracy and robustness of fault probability prediction.

[0051] Step 2: Calculate the satellite's position and relative motion state using the two lines of orbital data from the satellite and the orbital debris. Based on the spatial positional relationship and relative motion geometry between the satellite and the orbital debris at the rendezvous point, calculate the probability of a collision between the satellite and the orbital debris.

[0052] Calculating the probability of a satellite collision includes the following steps:

[0053] The orbital parameters of the satellite and orbital debris are determined, including orbital altitude, orbital inclination, and the right ascension of the ascending node used to determine the orbital plane orientation—basic orbital elements describing the orbital position and state. These orbital parameters can be obtained from the two-line orbit (TLE) data of the satellite and orbital debris. The probability density function of the velocity increment of the debris follows a normal distribution, and:

[0054]

[0055] In the formula: It represents the probability density; and The mean and standard deviation are normal distributions, given by existing fragmentation statistical models or determined by statistical fitting based on observed data of historical fragmentation events, and are considered known parameters in this invention; The velocity increment of the disintegrating fragments; Indicates the velocity increment of the fragment The decimal logarithm.

[0056] The accuracy analysis of satellite TLE data employs an internal coincidence method: The second TLE data point is selected as the reference orbit. The first and second TLE data points are predicted to the same epoch within the same time period. The difference between the position obtained from the first TLE data point and the position obtained from the second TLE data point is calculated, and this difference is statistically analyzed throughout the entire analysis period to obtain the accuracy of the first TLE data point within that period. The key to this method is finding the time period of the quasi-reference orbit that minimizes the error between the reference orbit and the actual orbit. The accuracy of forward and backward TLE predictions differs, primarily depending on the degree of overlap between the prediction time period and the quasi-reference orbit data time period used for TLE fitting. TLE data can be used to calculate collision probabilities and represent the position error covariance. , , Provide data sources.

[0057] Using a selected orbit prediction algorithm, such as the SGP4 / SDP4 orbit propagation algorithm widely used in TLE data, the position vector and velocity vector of the satellite and orbital debris in the inertial coordinate system are calculated at a given time based on the TLE orbit parameters of the satellite and orbital debris. The spatial distance between the satellite and the debris at a certain time is the modulus of the difference between their position vectors. When the spatial distance is less than a preset distance threshold, it is considered that there is a potential collision risk, and the collision probability needs to be further calculated.

[0058] During orbital rendezvous, the first step is to screen for hazardous targets. Given the large number of objects in space, calculating the orbits of every single one would be extremely computationally intensive. To reduce this, objects that could potentially collide with the target spacecraft are screened. This involves rapidly removing objects from the vast pool of space environment targets provided by the space monitoring and cataloging system that have no possibility of intersecting with the target orbit, thus identifying hazardous objects that could potentially collide with the target spacecraft or are within a certain distance (d). After eliminating targets with no collision risk, the collision probability of satellites and debris is calculated.

[0059] Probability of collision between satellite and orbital debris The probability of collision depends not only on the spatial relationship between the satellite and the space target at the moment of rendezvous, but also on factors such as the size of the two targets, their relative motion geometry, and position prediction errors (uncertainties). The calculation formula is:

[0060]

[0061]

[0062]

[0063]

[0064] in, The difference in height in the radial direction between the satellite and the orbital debris when they cross the line of intersection of the common orbital plane is obtained by projecting the predicted orbital positions of the satellite and the orbital debris in the radial direction at the moment of intersection and taking the difference. It is a unit vector along the direction of relative position; The distance between target 1 (satellite) and the intersection line when target 2 (orbital debris) reaches the intersection line. The angle between the track planes. This represents the relative position vector of the satellite and orbital debris at the moment of their encounter; Located in the plane of relative motion, and relative to the unit vector A vertical tangential unit vector; It is the unit normal vector of the plane of relative motion; The collision determination radius is used to characterize the minimum safe distance between two targets when a collision occurs.

[0065] Figure 3 This diagram illustrates the geometric relationship between the satellite and orbital debris in the encounter plane, and provides the relative position vectors. Angle between the track surfaces The projection relationship of the position error ellipsoid is used to intuitively illustrate the physical meaning of each geometric quantity in the above collision probability calculation formula. Among them, , , The projection radius of the position error ellipsoid along each coordinate axis is used to construct the position error covariance of the satellite and orbital debris at the moment of their encounter. This represents the target's relative position vector, whose direction is the same as the unit vector in the above formula. Consistent with the definition, these quantities are used together to determine the height difference within the meeting plane. ,distance And the variance required for collision probability calculation.

[0066] Step 3: Calculate the time weight of each time slot based on the static topology diagram to obtain the node importance of the satellite.

[0067] The satellite's node importance is determined using a multi-attribute topological importance algorithm with an improved importance contribution matrix. Initial importance is determined based on multiple attributes, including degree centrality (DC), betweenness centrality (BC), and local triangular centrality (LTC). The improved importance contribution matrix is ​​then used to measure the influence of neighboring nodes in the network. The main process is as follows: First, in the static topology graph of each time slot, the initial importance of a node is calculated using node degree centrality, betweenness centrality, and local triangular centrality. Second, an importance contribution matrix is ​​constructed based on the initial importance, proximity centrality, and distance between nodes to quantify the mutual influence between nodes. Next, the importance value of a node in each time slot is calculated. Time weights are determined based on the proportion of time slot duration to the entire orbital period, and the importance values ​​of each time slot are weighted and fused to obtain the comprehensive importance of the node throughout the entire period, i.e., the satellite's node importance. Finally, the change in importance over time is represented in the time-spread graph through node replicas.

[0068] Node importance is determined by considering both global and local features of the node, as well as the influence of its neighboring nodes. This method calculates node importance in detail in the static topology graph of each time slot through the following steps:

[0069] Step 3.1: Considering the local and global centers of satellite nodes, when constructing the initial importance and probability transfer matrix of nodes, indices such as degree centrality, betweenness centrality, and local triangle centrality are introduced simultaneously. The local connectivity characteristics and global positional characteristics of the nodes are incorporated into the importance calculation. The initial importance of the nodes... The calculation formula is:

[0070]

[0071] ,

[0072] in, , and Representing nodes respectively The center of betweenness, the center of degree, and the center of the local triangle. These represent the index weights for the betweenness center, degree center, and local triangle center, respectively. This indicates the total number of nodes in the satellite optical network.

[0073] node degree center The calculation is as follows:

[0074]

[0075] in, Represents a node The number of neighboring nodes, Representing the adjacency matrix The first in line, number The elements of a column. Degree centers describe the local characteristics of a node. The higher the degree center of a node in a network, the more first-order adjacent edges that node has, which means that the node occupies a central position in the network.

[0076] The betweenness center is defined as the proportion of shortest paths between any two nodes in a network that pass through a specific node. betweenness center The calculation formula is as follows:

[0077]

[0078] in, Represents a node To the node The number of all shortest paths, This indicates the nodes visited in the shortest path. The number of paths.

[0079] Local triangle centrality is a node-specific local triangle centrality proposed by combining the characteristics of neighboring nodes and the triangular structure in the network topology. Local triangle centrality The calculation formula is as follows:

[0080] ;

[0081] in, Represents a node The set of neighboring nodes, This refers to the sigmoid function, which can calculate the number of triangles. The value is mapped to the interval between 0 and 1. Represents a node neighboring nodes The degree (number of neighbors). This represents the adjustment factor, set to 1, indicating that the degree of a node and its neighbors are proportional to its importance. Represents a node and nodes The number of triangles formed between two nodes, that is, the number of adjacent nodes that these two nodes share. The calculation formula is:

[0082]

[0083] in, Represents a node The set of neighboring nodes, Represents a node The set of neighboring nodes.

[0084] Step 3.2: Calculate the importance contribution values ​​between nodes and construct the importance contribution matrix.

[0085] node near the center The calculation formula is:

[0086]

[0087] in, Represents a node and nodes The shortest distance.

[0088] The node importance contribution matrix is ​​a matrix representing the importance contribution values ​​between nodes, describing the degree of importance of nodes in the network. The calculation formula is:

[0089]

[0090] in, For nodes Approaching the center, Represents a node and nodes The shortest distance between them Represents a node The initial importance of the node. Other nodes in the network assign their own importance to the node. Proportional contribution to nodes .

[0091] Step 3.3: Calculate the importance value of the node in each time slot.

[0092] To accurately assess the importance of nodes, considering both the node's global importance and the contributions of its neighbors, the initial node importance is combined with an importance contribution matrix. In the time slot Importance value The calculation formula is:

[0093]

[0094] in, Represents a node The initial importance, Representing neighboring nodes Approaching the center, Represents a node and neighboring nodes The shortest distance between them.

[0095] The above calculations are performed in each time slot. Static topology graph Performed independently, through all This calculation is performed sequentially in each time slot to obtain the result for each node. Importance value in each time slot These values ​​form the basis for subsequent time-weighted fusion.

[0096] Step 3.4: Introduce time weights and perform cross-time slot importance fusion.

[0097] 1. Calculate time weights: to obtain the node's weights throughout the entire orbital period. The overall importance within a time slot requires time-weighting of the importance values ​​for each time slot. Let time slots be... The duration is Its time weight The calculation formula is:

[0098] .

[0099] 2. Weighted fusion: By weighting the nodes... Importance value of each time slot Time-weighted summation is used to obtain the nodes. The overall importance In other words, the importance of satellite nodes:

[0100]

[0101] Comprehensive node importance It is a single scalar value that comprehensively reflects the global importance of a node throughout the entire dynamic topology cycle, taking into account its own attributes, the influence of its neighbors, and the proportion of time.

[0102] Step 3.5: Visualize the time-spread graph

[0103] To intuitively describe the evolution of node importance over time, this invention provides a visual representation in a time-spread graph: in the time-spread graph, the same physical satellite in different time slots... In, it is represented as an independent node replica. This is used to characterize the network status of the satellite at different times. The calculated importance values ​​for each time slot are then used to... Assign the value to the corresponding node replica. Along the time axis, connect the node replicas of adjacent time slots of the same satellite.

[0104] The above process generates a three-dimensional spatiotemporal graph, such as Figure 4 As shown, by replicating and stacking the network topology of each time slot in chronological order, and marking satellite nodes of different risk levels with different colors according to the reliability evaluation results, we can intuitively observe the changes in the position of satellite nodes in the time spread map at each moment and the risk clusters to which they belong, providing an intuitive basis for avoiding high-risk nodes in subsequent route planning.

[0105] The time weight is the ratio of the duration of each time slot to the total orbital period, reflecting the relative importance of that time slot within the entire period. Node importance is obtained by weighted fusion of topological importance, where topological importance is calculated from the node's degree centrality, betweenness centrality, local triangle centrality, and neighbor contributions. The calculated node importance is used for reliability evaluation in steps 4-5.

[0106] Step 4: Preprocess the obtained satellite failure probability, collision probability, and node importance to eliminate differences between indicators, reduce the interference of outliers on data analysis, and improve model performance.

[0107] The calculated reliability-related index data, such as satellite component failure probability, collision probability, and node importance, are preprocessed. This involves normalizing each index to a unified dimension and direction, mapping them to the same numerical range, and truncating or smoothing obvious outliers to reduce the interference of extreme data on the results. This ensures that the processed data meets the requirements of the fuzzy comprehensive evaluation model for dimensionless and comparable inputs.

[0108] After the above steps, the satellite component failure probability, the collision probability between the satellite and orbital debris, and the node importance are obtained respectively. The vector composed of the three is used as the basic index for satellite reliability evaluation, and is preprocessed in a unified manner and input into the fuzzy comprehensive evaluation model.

[0109] Step 5: Use the preprocessed data as input to the fuzzy comprehensive evaluation model, use the fuzzy comprehensive evaluation method to evaluate the satellite reliability, and classify the satellite into three categories of satellite nodes: high-risk, general-risk, and low-risk based on the comprehensive evaluation results.

[0110] The fuzzy comprehensive evaluation model uses three indicators—satellite component failure probability, satellite-debris collision probability, and node importance—as evaluation factors, forming a factor set. It uses three satellite types—high-risk, moderate-risk, and low-risk—as the evaluation set. Based on the preprocessed indicator values, membership functions corresponding to different risk levels are constructed for each evaluation factor. The membership degree of each satellite at each risk level is calculated, forming a membership degree matrix. The indicator vector of a single satellite is used as input to the fuzzy comprehensive evaluation model. This method can evaluate multiple indicators and provide an overall assessment of satellite failure influencing factors using fuzzy data. Weight vectors are determined based on the relative importance of each evaluation factor in satellite reliability assessment. The membership degree matrix is ​​then weighted according to these weights to obtain the comprehensive membership degree for each risk level—essentially, fuzzy weighting is achieved through the weighted summation of the membership degrees of each indicator. Based on the obtained comprehensive membership degree, the risk level of the satellite is determined using the maximum membership degree principle, classifying satellites into high-risk, moderate-risk, and low-risk categories, thus completing the comprehensive evaluation of satellite reliability. High-risk nodes require special attention during route selection. For satellites with low reliability levels, priority protection should be given, and they can be avoided when designing routes for reliability-sensitive services.

[0111] Describing low-orbit satellites as , Represents the failure probability of satellite components. Indicates the probability of a collision between a satellite and debris. Indicating the importance of satellite nodes, among which The values ​​are , and These represent satellite reliability levels of high reliability, low reliability, and general reliability, respectively. Satellites with low reliability levels require close monitoring.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the reliability of satellite nodes in an F6G low-Earth orbit satellite optical network, characterized in that, The steps are as follows: Step 1: Initialize the low-Earth orbit satellite optical network topology, divide the orbital period into several time slots to obtain the static topology map corresponding to each time slot; collect satellite component status information and environmental data, and calculate the satellite failure probability; Step 2: Calculate the satellite's position and relative motion state using the two lines of orbital data from the satellite and the orbital debris. Based on the spatial positional relationship and relative motion geometry between the satellite and the orbital debris at the rendezvous point, calculate the probability of collision between the satellite and the orbital debris. Step 3: Calculate the time weight and importance value of each time slot based on the static topology diagram, and obtain the node importance of the satellite through weighted fusion; Step 4: Preprocess the obtained satellite failure probability, collision probability, and node importance; Step 5: Use the preprocessed data as input to the fuzzy comprehensive evaluation model to obtain the comprehensive evaluation results. Based on the comprehensive evaluation results, classify the satellites into three categories: high-risk, general-risk, and low-risk satellite nodes. The satellite component status information includes operating parameters and historical fault records, while the environmental data describes the satellite's operating environment and orbital characteristics. The method for calculating the node importance of the satellite is as follows: in the static topology graph of each time slot, the node degree centrality, betweenness centrality and local triangle centrality are selected to calculate the initial importance of the node; An importance contribution matrix is ​​constructed based on the initial importance, proximity centrality, and distance between nodes. The importance value of a node in each time slot is calculated using the importance contribution matrix. The time weight is determined based on the proportion of the time slot duration to the entire orbital period. The importance values ​​of each time slot are weighted and fused to obtain the comprehensive importance of the node in the entire orbital period, thus obtaining the node importance of the satellite.

2. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 1, characterized in that, The method for initializing the LEO satellite optical network topology is as follows: satellites are abstracted as nodes, inter-satellite links are abstracted as edges, and the LEO satellite optical network topology is abstracted as an undirected time-spreading graph. , where the set of nodes , This indicates the total number of nodes in the satellite optical network. Represents the set of inter-satellite links, and Represents a node and nodes Inter-satellite links; adjacency matrix of nodes , among which, element yes Represents a node and nodes Inter-satellite links exist, otherwise ; According to the time spread diagram theory, the satellite's orbital period The topology is divided into a series of snapshots. Where K represents the total number of time slots, and the k-th time slot is represented as... ,in, This represents the start time of the selected time period. This represents the end time of the selected time period; within the orbital period. Inside, in the Each time slot Static topology diagram of the built-in satellite optical network The entire orbital period The network topology diagram is represented as .

3. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 2, characterized in that, The operating parameters include the operational status information of the satellite node's computing resources, storage resources, and link bandwidth; historical fault records are generated during the long-term operation of the satellite system. The historical fault data of the historical fault record is processed for missing values ​​and outliers. The normalization method is used to map the working parameters of different dimensions to a comparable numerical range. Then, the various working parameters and historical fault data are organized into a satellite element state characteristic sequence according to the time sequence and the corresponding fault markers. Satellite failures can be categorized into early failure periods, random failure periods, and attrition failure periods.

4. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 3, characterized in that, The method for calculating the satellite failure probability is as follows: the satellite component state feature sequence, which is composed of preprocessed satellite component state information, is input into the failure probability model to obtain the failure probability of the satellite component.

5. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 4, characterized in that, The fault probability model employs a Long Short-Term Memory (LSTM) network with an attention mechanism, comprising an input layer, a CNN layer, an LSTM layer, an attention layer, and an output layer connected in sequence. The input layer receives the satellite component state feature sequence; the CNN layer performs local convolution operations on the input satellite component state feature sequence to extract local features at different times and scales; the LSTM layer models the dependencies of the feature sequence over time; the attention layer includes an attention mechanism and a fully connected layer. The attention mechanism calculates weight coefficients based on the importance of different time steps or features to fault determination, weighting the output features of the LSTM layer; the fully connected layer performs a linear transformation on the weighted features; and the output layer maps the result of the attention mechanism to the corresponding fault probability and outputs it.

6. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to any one of claims 2-5, characterized in that, The probability density function of the velocity increment of the orbital debris follows a normal distribution; The internal coincidence method is used to analyze two lines of orbit data to find the time period of the quasi-reference orbit, so that the error between the reference orbit and the actual orbit is minimized during that time period; Using the selected orbit prediction algorithm, the position vector and velocity vector in the inertial coordinate system are calculated at a given time based on the two lines of orbit data of the satellite and the orbital debris. When the spatial distance between the satellite and the orbital debris is less than a pre-set distance threshold at a certain time, it is considered that there is a potential collision risk between the satellite and the orbital debris, and the collision probability is calculated.

7. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 6, characterized in that, The collision probability for: ； ； ； ； in, The radial height difference between the satellite and orbital debris when they cross the line of intersection of their common orbital plane; It is a unit vector along the direction of relative position; The distance between the satellite and the intersection line when the orbital debris arrives. The angle between the track surfaces; This represents the relative position vector of the satellite and orbital debris at the moment of their encounter; Located in the plane of relative motion, and relative to the unit vector A vertical tangential unit vector; It is the unit normal vector of the plane of relative motion; The collision detection radius, Indicates the velocity increment of the fragment The decimal logarithm.

8. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 7, characterized in that, The method for determining the importance value of the computing node in each time slot is as follows: In each time slot Static topology graph Taking into account the global importance of a node and the contributions of its neighbors, the nodes are... initial importance Combined with the importance contribution matrix, the i-th node is obtained. In the time slot Importance value for: ； The time weight for: ； For all The satellite's nodal importance is obtained by time-weighted summation of the importance values ​​of each time slot: ； in, Represents a node and neighboring nodes The shortest distance between them Representing neighboring nodes Approaching the center, Neighboring nodes The initial importance; For time slots The duration.

9. The satellite node reliability evaluation method in the F6G low-Earth orbit satellite optical network according to claim 8, characterized in that, The preprocessing in step 4 involves normalizing the data on satellite failure probability, collision probability, and node importance to a unified dimension and direction, mapping them to the same numerical range, and truncating or smoothing outliers. The preprocessed data on satellite failure probability, collision probability, and node importance are then used as a vector input into the fuzzy comprehensive evaluation model. The fuzzy comprehensive evaluation model uses satellite failure probability, collision probability, and node importance as evaluation factors to form a factor set; and uses three types of satellites—high-risk, general-risk, and low-risk—as the evaluation set. Based on the preprocessed satellite failure probability, collision probability, and node importance index values, a membership function corresponding to different risk levels is constructed for each evaluation factor. The membership degree of each satellite under each risk level is calculated to form a membership matrix. The weight vector is determined based on the relative importance of each evaluation factor in the satellite reliability assessment. The membership matrix is ​​then weighted according to the weights to obtain the comprehensive membership degree of each risk level. Based on the obtained comprehensive membership degree, the risk level of the satellite is determined using the maximum membership degree principle, and the satellite is divided into three categories: high risk, general risk, and low risk.

Images