Data construction method and system based on satellite-ground cooperative computing

Abstract

The embodiment of the invention discloses a data construction method and system based on satellite-ground cooperative computing, and belongs to the technical field of data processing. According to the embodiment of the invention, the condition of the satellite-ground collaborative network matrix can be comprehensively mastered by acquiring the ground side sensing data and the satellite side state data. Data features can be effectively extracted by mining coding vectors, data redundancy is reduced, and a precise basis is provided for subsequent integration. The first vector integration and the second vector integration adopt different interaction strategies, and satellite-ground data can be deeply fused, so that the obtained satellite-ground network embedded integrated vector can fully reflect the complex relationship of the satellite-ground collaborative network. Decision information obtained by judging the data construction situation according to the vector is favorable for optimizing data construction of a satellite-ground collaborative network matrix, improving rationality of network resource allocation, enhancing the ability of the network to cope with complex environment changes, and ensuring efficient and stable operation of the satellite-ground collaborative network.

Description

Technical Field

[0001] The embodiments of the present application belong to the field of data processing technology, and specifically relate to a data construction method and system based on satellite-ground collaborative computing. Background Art

[0002] In the field of satellite-ground collaborative networks, with the continuous development of satellite technology and ground network technology, how to effectively manage and optimize the data construction of the satellite-ground collaborative network matrix has become a technical problem that needs to be solved urgently. Traditional network management methods are difficult to fully take into account the status information of ground network equipment and satellites, resulting in insufficient data integration and failure to accurately reflect the overall situation of the satellite-ground collaborative network. For example, in terms of resource allocation, the lack of consideration of the overall situation of the satellite and the ground can easily lead to waste of resources or insufficient resources in certain areas. At the same time, traditional methods cannot make timely adaptive adjustments to the dynamic changes of the network. Summary of the invention

[0003] The embodiments of the present application provide a data construction method and system based on satellite-ground collaborative computing, which can solve the technical problems involved in the above-mentioned background technology.

[0004] An embodiment of the present application provides a data construction method based on satellite-ground collaborative computing, which is applied to a satellite-ground collaborative data construction system, the method comprising: obtaining ground-side perception data and multiple groups of satellite-side status data corresponding to a satellite-ground collaborative network matrix, wherein the ground-side perception data is ground network equipment monitoring data associated with the satellite-ground collaborative network matrix; mining ground network embedded coding vectors corresponding to the ground-side perception data, and mining satellite network embedded coding vectors corresponding to the multiple groups of satellite-side status data; performing a first vector integration on multiple satellite network embedded coding vectors and the ground network embedded coding vectors to obtain first satellite-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors; performing a second vector integration on the first satellite-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors to obtain a second satellite-ground network embedded integration vector corresponding to the satellite-ground collaborative network matrix, wherein the vector feature interaction strategies of the first vector integration and the second vector integration are different; performing data construction situation judgment on the satellite-ground collaborative network matrix according to the second satellite-ground network embedded integration vector to obtain construction situation decision information corresponding to the satellite-ground collaborative network matrix.

[0005] An embodiment of the present application provides a satellite-ground collaborative data construction system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0006] An embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the above method are implemented.

[0007] The embodiments of the present application can fully grasp the status of the satellite-ground collaborative network matrix by acquiring ground-side perception data and satellite-side status data. Mining coded vectors can effectively extract data features, reduce data redundancy, and provide a precise basis for subsequent integration. The first vector integration and the second vector integration adopt different interaction strategies to deeply integrate satellite-ground data, so that the obtained satellite-ground network embedded integration vector can fully reflect the complex relationship of the satellite-ground collaborative network. The decision information obtained by judging the data construction situation based on the vector helps to optimize the data construction of the satellite-ground collaborative network matrix, improve the rationality of network resource allocation, enhance the network's ability to cope with complex environmental changes, and ensure the efficient and stable operation of the satellite-ground collaborative network. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A flowchart of a data construction method based on satellite-ground collaborative computing provided in an embodiment of the present application.

[0009] Figure 2 A schematic diagram of the structure of a satellite-ground collaborative data construction system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of them. Based on the embodiments in the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.

[0011] The terms "first", "second", etc. in the embodiments of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally a class, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the embodiments of the present application, "and / or" represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0012] Figure 1 A data construction method based on satellite-ground collaborative computing is shown, which is applied to a satellite-ground collaborative data construction system. The method includes the following steps 101 to 105.

[0013] Step 101: The satellite-ground collaborative data construction system obtains ground-side perception data corresponding to the satellite-ground collaborative network matrix and multiple groups of satellite-side status data, wherein the ground-side perception data is ground network equipment monitoring data associated with the satellite-ground collaborative network matrix.

[0014] Step 102: The satellite-ground collaborative data construction system mines the ground network embedded coding vectors corresponding to the ground-side perception data, and mines the satellite network embedded coding vectors corresponding to the multiple groups of satellite-side status data.

[0015] Step 103: The satellite-ground collaborative data construction system performs first vector integration on the multiple satellite network embedded coding vectors and the ground network embedded coding vector respectively to obtain first satellite-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors respectively.

[0016] Step 104: The satellite-ground collaborative data construction system performs a second vector integration on the first satellite-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors to obtain a second satellite-ground network embedded integration vector corresponding to the satellite-ground collaborative network matrix. There is a difference in the vector feature interaction strategies between the first vector integration and the second vector integration.

[0017] Step 105: The satellite-ground collaborative data construction system performs data construction situation judgment on the satellite-ground collaborative network matrix according to the second satellite-ground network embedding integrated vector to obtain construction situation decision information corresponding to the satellite-ground collaborative network matrix.

[0018] In the embodiment of the present application, the satellite-ground collaborative network matrix is ​​a network architecture that organically combines the satellite network and the ground network. In this network architecture, there is a specific connection relationship, communication link, and interaction mode between the nodes of the satellite network and the ground network. It is the basic framework of the entire satellite-ground collaborative data construction system, providing structural support for the circulation, interaction, and related operations of data.

[0019] First, in step 101, the satellite-ground collaborative data construction system starts to acquire relevant data. This involves ground-side perception data and multiple sets of satellite-side status data. Ground-side perception data is ground network equipment monitoring data associated with the satellite-ground collaborative network matrix. For example, ground network equipment monitoring data may include operating status data of ground base stations, such as the temperature of the base station (for example, the normal operating temperature range is between 10-35 degrees Celsius), power consumption (1-5 kilowatts of electricity may be consumed per hour, and different base stations vary due to equipment scale and business volume), the number of user connections within the signal coverage range (thousands of user connections may be reached during busy hours), etc. These data reflect the actual operation of ground network equipment in the satellite-ground collaborative network matrix. Satellite-side status data may include satellite orbital position information (for example, the orbital altitude of a satellite is about 500 kilometers, the orbital inclination is 45 degrees, etc.), the working status of each sensor on the satellite (such as the sampling frequency of a sensor is 10 times per second, etc.), the energy reserve of the satellite (for example, the initial energy reserve of the satellite can support its work for 5 years, and it is gradually consumed as it is used), and other aspects of data.

[0020] Then enter step 102, the satellite-ground collaborative data construction system performs a mining operation on the acquired data. For the ground-side perception data, the system mines the corresponding ground network embedded coding vector. This coding vector is a representation form after feature extraction and encoding of the ground-side perception data. For example, the temperature, power consumption, number of user connections and other data of the ground base station can be encoded according to a certain algorithm. For example, the temperature value is encoded as an interval of every 5 degrees, the power consumption is encoded as an interval of every 1 kilowatt, and the number of user connections is encoded as an interval of every 1000. In this way, the complex state information of the ground base station can be converted into a specific coding vector. Similarly, for the satellite-side status data, the system mines the satellite network embedded coding vector corresponding to each group of satellite-side status data. Taking the orbital position of the satellite, the working status of the sensor and the energy reserve as an example, a similar interval division method can also be used for encoding. For example, the orbital height is encoded as an interval of every 100 kilometers, the sensor sampling frequency is encoded as an interval of every 5 times per second, and the energy reserve is encoded as an interval according to the annual consumption ratio, thereby obtaining a satellite network embedded coding vector.

[0021] Then comes step 103, where the satellite-ground collaborative data construction system performs first vector integration on multiple satellite network embedded coding vectors and ground network embedded coding vectors, respectively, to obtain first satellite-ground network embedded integrated vectors corresponding to multiple satellite network embedded coding vectors. The vector integration here is an operation of fusing coding vectors from different sources. For example, there are three groups of satellite network embedded coding vectors, namely, coding vector V1 corresponding to satellite 1, coding vector V2 corresponding to satellite 2, coding vector V3 corresponding to satellite 3, and ground network embedded coding vector G. When performing the first vector integration, a weighted summation method may be used (this is just an example method, and a more complex algorithm may be used in practice). For example, a weight of 0.3 is assigned to V1, a weight of 0.3 is assigned to V2, a weight of 0.3 is assigned to V3, and a weight of 0.1 is assigned to G. Then, the first satellite-ground network embedded integrated vector I1 corresponding to satellite 1 is 0.3V1+0.1G, the first satellite-ground network embedded integrated vector I2 corresponding to satellite 2 is 0.3V2+0.1G, and the first satellite-ground network embedded integrated vector I3 corresponding to satellite 3 is 0.3V3+0.1G. This integration method preliminarily integrates the characteristics of satellite networks and ground networks, forming the first satellite-to-ground network embedded integration vector for each satellite.

[0022] Then, in step 104, the satellite-ground collaborative data construction system performs a second vector integration on the first satellite-ground network embedded integrated vectors corresponding to the multiple satellite network embedded coding vectors, and obtains the second satellite-ground network embedded integrated vector corresponding to the satellite-ground collaborative network matrix. Here, there is a difference in the vector feature interaction strategy between the first vector integration and the second vector integration. In the second vector integration process, a weighting method or fusion algorithm different from the first vector integration may be used. For example, continuing to take the first satellite-ground network embedded integrated vectors I1, I2, and I3 corresponding to the above-mentioned satellites 1, 2, and 3 as an example, in the second vector integration, a more complex nonlinear fusion algorithm may be used. For example, according to the overall structural characteristics of the satellite-ground collaborative network matrix and the importance distribution of the data, I1, I2, and I3 are recombined and fused. It is possible to amplify some features in I1 (for example, multiply by 2), reduce some features in I2 (for example, multiply by 0.5), and adjust some features in I3 (for example, take its square root), and then sum these adjusted features or perform other complex combination operations to finally obtain the second satellite-ground network embedded integrated vector J corresponding to the satellite-ground collaborative network matrix.

[0023] Finally, in step 105, the satellite-ground collaborative data construction system performs data construction situation discrimination on the satellite-ground collaborative network matrix based on the second satellite-ground network embedded integrated vector, and obtains the construction situation decision information corresponding to the satellite-ground collaborative network matrix. The data construction situation discrimination here is to judge the current state and development trend of the satellite-ground collaborative network matrix based on the comprehensive feature information contained in the second satellite-ground network embedded integrated vector. For example, if some characteristic values ​​in the second satellite-ground network embedded integrated vector J indicate that the orbital position change rate of the satellite is large (exceeding a certain preset threshold, such as this threshold is an orbital angle change of 0.1 degrees per day), and the number of user connections of the ground base station increases rapidly (increasing by more than 100 user connections per hour), then the system may judge that the satellite-ground collaborative network matrix is ​​in a rapidly changing situation, and the data construction strategy needs to be adjusted, such as increasing the data transmission bandwidth between the satellite and the ground base station (from the original 10Mbps to 20Mbps) or optimizing the routing selection of data transmission. This construction situation decision information can provide a decision basis for the data construction of the satellite-ground collaborative network matrix, ensuring that the entire satellite-ground collaborative data construction system can perform effective data management and network optimization according to actual conditions.

[0024] Through the above series of steps, the satellite-ground collaborative data construction system comprehensively performs data construction-related operations on the satellite-ground collaborative network matrix, from data acquisition, coding vector mining, vector integration to final situation identification, providing effective technical support for the stable operation, optimized development and efficient data management of the satellite-ground collaborative network.

[0025] Continuing to elaborate on this process, the richness and accuracy of the data obtained in step 101 are crucial to the implementation of the entire invention application. In addition to the typical data types mentioned above, ground-side perception data may also include link delays between ground network devices (under normal circumstances, link delays may be between 1-10 milliseconds, and different links vary due to distance and device performance), data packet loss rates (usually expected to remain below 1%) and other information. Satellite-side status data may also cover the occupancy of satellite communication frequency bands (for example, the utilization rate of a certain frequency band is about 30%), the remaining capacity of data storage devices on the satellite (for example, the total capacity is 1TB, and the remaining capacity is 0.3TB), etc. The accurate acquisition of these data provides a comprehensive foundation for subsequent operations.

[0026] When mining the coding vector in step 102, in order to improve the effectiveness of the coding, a variety of data mining techniques can be used. For example, for the mining of ground network embedded coding vectors, a cluster analysis method can be used to classify ground network devices with similar characteristics, and then encode according to the classification results. For example, ground base stations are divided into three categories of high, medium and low according to the population density of the coverage area, and are encoded with 1, 2, and 3 respectively, and then combined with other data features (such as temperature, power consumption, etc.) to form a complete ground network embedded coding vector. For the mining of satellite network embedded coding vectors, it may be necessary to consider the particularity of the satellite, such as the orbit type of the satellite (geostationary orbit, low earth orbit, etc.). Satellites of different orbit types may have different weights or encoding rules when encoding to reflect their different roles in the satellite-ground collaborative network matrix.

[0027] In the first vector integration process of step 103, in addition to the example method of weighted summation, an integration method based on a neural network can also be used. For example, a feedforward neural network is constructed, and the satellite network embedding coding vector and the ground network embedding coding vector are used as neuron inputs of the input layer. The data is processed through the hidden layer of the neural network (for example, the hidden layer has 3 neurons, and each neuron uses a different activation function, such as sigmoid, tanh, and rectified linear unit function (ReLU)), and finally the first satellite-ground network embedding integration vector is obtained in the output layer. This method can better mine the nonlinear relationship between data and improve the effect of vector integration.

[0028] In the second vector integration of step 104, the difference in the vector feature interaction strategy is more obvious. An optimization strategy based on a genetic algorithm can be used to determine the interaction mode between different vector features. For example, each feature in the first satellite-to-ground network embedding integrated vector is regarded as a gene in a genetic algorithm, and the optimal feature combination mode is found through the selection, crossover and mutation operations of the genetic algorithm to construct the second satellite-to-ground network embedding integrated vector. This method can adaptively adjust the vector integration strategy according to the complex environment and requirements of the satellite-to-ground collaborative network matrix, thereby improving the flexibility and adaptability of the system.

[0029] In step 105, the second satellite-to-ground network embedded integrated vector based on the data construction situation judgment contains rich information. In addition to the information such as the satellite orbital position and the number of ground base station user connections mentioned above, it may also include the bit error rate of data transmission between the satellite and the ground network (for example, the bit error rate should be lower than 0.01% under normal circumstances), the overall load balance of the network (evaluated by calculating the load ratio of different nodes), etc. When judging the situation based on this information, a fuzzy logic method can be used. For example, the value of each feature is mapped to different fuzzy sets (such as "low", "medium", "high", etc.), and then judged according to pre-defined fuzzy rules (such as "if the satellite orbital position change rate is high and the number of ground base station user connections is high, then the situation is rapid expansion") to obtain accurate construction situation decision information. This fuzzy logic method can handle some imprecise and uncertain information, and better adapt to the complex and changeable actual situation of the satellite-to-ground collaborative network matrix.

[0030] During the implementation of the entire invention application, the satellite-ground collaborative data construction system manages and optimizes the data construction of the satellite-ground collaborative network matrix from multiple angles through carefully designed steps, ensuring that the system can operate effectively and make reasonable decisions in different operating environments.

[0031] Further elaborating on more details in each step, in step 101, the process of the satellite-ground collaborative data construction system acquiring the ground-side perception data corresponding to the satellite-ground collaborative network matrix and multiple sets of satellite-side status data is a complex process of multi-source data acquisition. For ground-side perception data, the system needs to establish a reliable communication connection with many ground network devices to obtain data. For example, ground network devices may be distributed in different geographical locations, from urban centers to remote mountainous areas, and the system needs to use multiple communication protocols (such as TCP / IP protocols, etc.) to ensure accurate data transmission. When acquiring base station temperature data, the analog signal may be converted into a digital signal by a temperature sensor and sent to the satellite-ground collaborative data construction system through the communication module inside the base station. For satellite-side status data, the system needs to communicate with the satellite. The satellite sends its own status data to the ground receiving station through a satellite communication link, and then forwards it to the satellite-ground collaborative data construction system. In this process, the transmission of data may be affected by various factors, such as atmospheric interference, the influence of solar activity, etc. In order to ensure the integrity of the data, data verification and error correction technologies such as cyclic redundancy check (CRC) may be used.

[0032] In step 102, the process of mining the coding vector involves a deep understanding of the data and feature extraction. For the mining of ground network embedded coding vectors, in addition to the cluster analysis and interval partition coding methods mentioned above, principal component analysis (PCA) technology can also be used. PCA can help identify the main components in the ground-side perception data, reduce the high-dimensional data to a low-dimensional space, and then encode it in the low-dimensional space. For example, when processing multiple state data (such as temperature, power consumption, number of user connections, link delay, etc.) containing multiple ground base stations, PCA can find several principal components that best represent the changes in these data, and then encode them according to the values ​​of these principal components. For the mining of satellite network embedded coding vectors, considering the spatiotemporal characteristics of satellite data, wavelet analysis technology can be used. Wavelet analysis can analyze the characteristics of satellite data at different scales and time frequencies, such as the long-term and short-term change trends of satellite orbital positions, periodic changes in sensor working conditions, etc., and then encode according to the results of wavelet analysis, so as to obtain a coding vector that more accurately reflects the satellite side state data.

[0033] In the first vector integration of step 103, in addition to the weighted summation and neural network methods mentioned above, the idea of ​​support vector machine (SVM) can also be used to construct an integration model. The satellite network embedding coding vector and the ground network embedding coding vector are regarded as different feature vectors, and an optimal hyperplane is found through SVM to integrate these vectors. In this process, it is necessary to determine the kernel function of SVM (such as linear kernel function, Gaussian kernel function, etc.) according to the distribution characteristics of the data. For example, if the data presents linearly separable features in the feature space, then a linear kernel function can be selected; if the data presents a nonlinear distribution, a Gaussian kernel function may be selected. This vector integration method based on SVM can improve the generalization ability of vector integration to a certain extent, so that the first satellite-to-ground network embedding integration vector obtained is more representative.

[0034] In the second vector integration of step 104, the design of the vector feature interaction strategy is based on the consideration of the overall performance optimization of the satellite-ground collaborative network matrix. In addition to the optimization strategy based on the genetic algorithm, a method based on particle swarm optimization (PSO) can also be used. PSO regards the features of each first satellite-ground network embedded in the integrated vector as the position and velocity of the particle, and searches for the optimal feature combination through the movement of the particle swarm in the search space. In the PSO algorithm, particles continuously adjust their positions and velocities based on their own experience and the experience of the group, and finally find the optimal solution. This method can quickly converge to a better vector feature interaction strategy and improve the efficiency and effect of the second vector integration.

[0035] In step 105, when the data construction situation of the satellite-ground collaborative network matrix is ​​judged according to the second satellite-ground network embedded integrated vector, in addition to the fuzzy logic method, a decision tree model can also be used. Each feature in the second satellite-ground network embedded integrated vector is used as the input node of the decision tree, and a decision tree model is constructed by learning a large amount of historical data. For example, based on the past satellite-ground collaborative network matrix operation data, a decision tree is constructed, in which the internal nodes represent different feature judgment conditions (such as whether the satellite orbit position change rate is greater than a certain threshold), and the leaf nodes represent different situation results (such as stable situation, expansion situation, contraction situation, etc.). When the new second satellite-ground network embedded integrated vector is input, the corresponding construction situation decision information is obtained through the decision path of the decision tree. This decision tree model has the advantages of being intuitive, easy to understand and explain, and can provide a clear decision basis for the data construction of the satellite-ground collaborative network matrix.

[0036] The satellite-ground collaborative data construction system can more accurately manage the data construction of the satellite-ground collaborative network matrix by continuously and deeply optimizing the technical means and methods in each step, improve the operating efficiency and stability of the entire satellite-ground collaborative network, and provide a solid technical guarantee for future satellite-ground collaborative communications and data interaction.

[0037] In summary, the embodiments of the present application can fully grasp the status of the satellite-ground collaborative network matrix by acquiring ground-side perception data and satellite-side status data. Mining coding vectors can effectively extract data features, reduce data redundancy, and provide a precise foundation for subsequent integration. The first vector integration and the second vector integration adopt different interaction strategies to deeply integrate satellite-ground data, so that the obtained satellite-ground network embedded integration vector can fully reflect the complex relationship of the satellite-ground collaborative network. The decision information obtained by judging the data construction situation based on the vector helps to optimize the data construction of the satellite-ground collaborative network matrix, improve the rationality of network resource allocation, enhance the network's ability to cope with complex environmental changes, and ensure the efficient and stable operation of the satellite-ground collaborative network.

[0038] In a preferred embodiment, performing a first vector integration on the multiple satellite network embedded coding vectors and the ground network embedded coding vectors to obtain first satellite-to-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors, respectively, includes: performing network state attribute fusion on the multiple satellite network embedded coding vectors and the ground network embedded coding vectors, respectively, to obtain first satellite-to-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors, respectively.

[0039] Optionally, before fusing the network state attributes of the multiple satellite network embedded coding vectors with the ground network embedded coding vectors to obtain the first satellite-to-ground network embedded integrated vectors corresponding to the multiple satellite network embedded coding vectors, the method further includes: performing vector pooling processing on the multiple satellite network embedded coding vectors to obtain satellite network embedded pooling vectors corresponding to the multiple satellite network embedded coding vectors, wherein the vector pooling processing is used to compress the feature size corresponding to the satellite network embedded coding vectors. Based on this, fusing the network state attributes of the multiple satellite network embedded coding vectors with the ground network embedded coding vectors to obtain the first satellite-to-ground network embedded integrated vectors corresponding to the multiple satellite network embedded coding vectors includes: fusing the satellite network embedded pooling vectors corresponding to the multiple satellite network embedded coding vectors with the ground network embedded coding vectors to obtain the first satellite-to-ground network embedded integrated vectors corresponding to the multiple satellite network embedded coding vectors.

[0040] In the above preferred embodiment, the core of the process of performing first vector integration on multiple satellite network embedding code vectors and ground network embedding code vectors to obtain first satellite-ground network embedding integration vectors corresponding to the multiple satellite network embedding code vectors is the fusion of network state attributes.

[0041] First, before the network state attribute fusion is performed, there is an operation of performing vector pooling on multiple satellite network embedded coding vectors. The purpose of vector pooling is to compress the feature size corresponding to the satellite network embedded coding vector, which is an important step to optimize the data structure and improve processing efficiency.

[0042] Take a satellite network embedding code vector as an example, for example, it contains 100 feature dimensions. These feature dimensions cover multiple aspects of the satellite's status information. For example, the satellite's orbital position information may occupy 5 dimensions, which are used to represent the three-dimensional coordinates of the orbit and the two components of the orbital velocity. The satellite may be equipped with 20 sensors, and the working status of each sensor is represented by 2 dimensions, with a total of 40 dimensions used to describe the sensor status. The satellite's energy reserve status uses 5 dimensions to represent the status of different energy reserve modules. In addition, the occupancy of the satellite communication frequency band uses 10 dimensions to represent the occupancy ratio of different frequency bands, etc. There are also some other dimensions used to represent other status information of the satellite, thus forming a vector of these 100 feature dimensions.

[0043] Through vector pooling, this vector of 100 feature dimensions can be compressed to, for example, a satellite network embedding pooling vector of 30 feature dimensions. This compression process is carefully designed and is based on an in-depth analysis and screening of the importance of features in the satellite network embedding coding vector. For example, in terms of the sensor state representation of the satellite, there may be some sensor state dimensions that are relatively minor in the current network state. For example, some sensor state dimensions that are only used for auxiliary monitoring and have little impact on the situation judgment of the satellite-ground collaborative network matrix data construction can be merged or discarded. For example, some sensor state dimensions that monitor micro-environmental changes inside the satellite are less important in the overall network resource allocation and data construction situation judgment, so they can be discarded, thereby retaining the most critical feature dimensions for the situation judgment of the satellite-ground collaborative network matrix data construction, such as sensor state dimensions that directly affect communication and data transmission.

[0044] Then, the network state attribute fusion operation is performed. Taking the previously obtained satellite network embedding pooling vector and ground network embedding coding vector as an example, the network state attribute fusion is an operation that deeply combines the satellite network and the ground network at the network state attribute level.

[0045] The ground network embedded coding vector also contains rich information. Taking a ground base station as an example, its ground network embedded coding vector contains information such as the base station temperature, power consumption, number of user connections within the signal coverage area, and link status between the ground base station and other base stations. Specifically, the temperature of the base station may use three dimensions to represent the temperature of different parts, for example, one dimension represents the temperature of the core part of the base station equipment, one dimension represents the temperature of the base station shell, and another dimension represents the temperature near the heat dissipation components of the base station. Power consumption uses two dimensions to represent the power consumption of different modules, such as one dimension represents the power consumption of the signal transmission module, and the other dimension represents the power consumption of the data processing module inside the base station. The number of user connections within the signal coverage area uses five dimensions to represent the distribution of the number of user connections in different time periods, for example, one dimension represents the number of user connections during the morning peak period, one dimension represents the number of user connections during the noon period, and so on. The link status between the ground base station and other base stations uses 10 dimensions to represent the bandwidth, delay, packet loss rate, etc. of the link, such as 5 dimensions are used to represent the bandwidth of the link with the adjacent base station, 3 dimensions represent the link delay, and 2 dimensions represent the packet loss rate.

[0046] When performing network status attribute fusion, the satellite orbit position information embedded in the pooled vector of the satellite network is associated and fused with the base station geographic location information embedded in the coding vector of the ground network. For example, the satellite's orbit altitude is 500 kilometers and the orbit inclination is 45 degrees. Its orbit can cover the area where a specific ground base station is located. For example, the base station is located at a longitude of 120 degrees and a latitude of 30 degrees. In the first satellite-to-ground network embedded integrated vector after fusion, the attributes of this coverage relationship will be reflected, which may be represented by a specific coding value or attribute identifier. This coding value or attribute identifier can be a value or symbol generated according to a pre-set rule. For example, if the satellite orbit completely covers the area where the base station is located, it is represented by a value of 1, and if it is partially covered, it is represented by 0.5, etc.

[0047] For another example, the energy reserve of the satellite is integrated with the power consumption of the ground base station. For example, the initial energy reserve of the satellite can support its operation for 5 years, and it is expected to support 2 years with the current remaining energy, while the power consumption of the ground base station fluctuates between 1-5 kilowatts per hour. When the satellite energy reserve is low, the ground base station can appropriately adjust its power consumption strategy to reduce its dependence on the overall resources of the satellite-ground collaborative network. This relationship will be reflected in the fused vector as an attribute that reflects the resource coordination strategy. For example, a range of attribute values ​​can be set. If the satellite energy reserve is at a low level and the ground base station power consumption is high, this attribute value may be in a higher range, such as 0.8-1, indicating that the power consumption strategy needs to be adjusted as soon as possible; if the satellite energy reserve is sufficient and the ground base station power consumption is within the normal range, this attribute value may be in the range of 0.2-0.5, indicating that the current resource status is good and no adjustment is required.

[0048] Attribute fusion is also performed for the working status of sensors on the satellite and the number of user connections within the signal coverage of the ground base station. For example, satellite sensors can detect some environmental information of the ground area, such as atmospheric humidity fluctuating between 30% and 80%, and the number of user connections of the ground base station may have different changing trends under different atmospheric humidity environments. In the fusion process, the correlation between this environmental factor and the number of user connections will be integrated into the first satellite-to-ground network embedded integrated vector. For example, when the atmospheric humidity is in the range of 30%-50%, a certain attribute value in the first satellite-to-ground network embedded integrated vector obtained by fusion will indicate that the ground base station should expect that the number of user connections may increase or decrease, so as to make resource allocation adjustments in advance. For example, if the attribute value shows that the number of user connections may increase, the ground base station can increase the signal transmission power or optimize the signal allocation strategy in advance to meet the possible increase in user needs.

[0049] Similarly, the occupancy of the satellite communication frequency band and the link status between the ground base station will also be deeply integrated with the network status attributes. For example, the occupancy rate of the satellite communication frequency band is about 30%, the link bandwidth between the ground base station and the adjacent base station is 10Mbps, the delay is 5 milliseconds, and the packet loss rate is 0.5%. If the occupancy of the satellite communication frequency band is relatively tight, the requirements for adjusting the link status of the ground base station will be reflected in the fused vector. For example, the ground base station may need to adjust the frequency band of its data transmission or optimize the link routing to avoid conflicts with satellite communications or make better use of satellite communication resources. If the occupancy rate of the satellite communication frequency band exceeds 50%, a certain attribute value in the fused vector may trigger the ground base station to adjust the link strategy, such as adjusting the link bandwidth from 10Mbps to 8Mbps to reduce potential interference with the satellite communication frequency band, while optimizing the internal data processing process and reducing the packet loss rate.

[0050] This fusion is all-round and multi-dimensional. By fusing the satellite network embedding pooling vector and the ground network embedding coding vector for network status attributes, the first satellite-ground network embedding integrated vector corresponding to the multiple satellite network embedding coding vectors can fully reflect the comprehensive status attributes of the satellite-ground collaborative network. It is no longer a simple piecing together of satellite network and ground network data, but an organic combination at the network status attribute level, deeply exploring the inherent connection between the two, and reflecting this connection in an orderly manner in the integrated vector.

[0051] The above technical solution provides a more accurate and effective data foundation for the construction of the entire satellite-ground collaborative network data. In the subsequent second vector integration and the final satellite-ground collaborative network matrix data construction situation judgment process, based on the first satellite-ground network embedded integrated vector that accurately reflects the comprehensive state attributes of the satellite-ground collaborative network, the operation status of the satellite-ground collaborative network can be more accurately grasped, more reasonable resource allocation decisions can be made, the overall architecture and operation efficiency of the satellite-ground collaborative network can be optimized, and the ability of the satellite-ground collaborative network to cope with complex environmental changes can be improved, thereby enhancing the stability and adaptability of the satellite-ground collaborative network, ensuring that the satellite-ground collaborative network can operate efficiently and stably in various complex working scenarios.

[0052] It can be seen that based on the above technical solution, vector pooling processing reduces data redundancy and improves the speed and efficiency of data processing by compressing the characteristic size of the satellite network embedded coding vector. Numerically, it is reflected in compressing the original vector of 100 characteristic dimensions to 30 dimensions, which greatly reduces the amount of data. Secondly, the fusion of network state attributes comprehensively and deeply considers the relationship between various state attributes of satellite networks and ground networks. For example, the fusion of satellite orbit and base station position, satellite energy and base station power consumption, satellite sensor status and number of base station user connections, satellite communication frequency band and base station link status. This enables the first satellite-to-ground network embedded integrated vector to accurately and comprehensively reflect the comprehensive state of the satellite-to-ground collaborative network. Subsequent operations based on such vectors will help to more accurately grasp the operating status of the satellite-to-ground collaborative network, provide a reliable basis for resource allocation, network architecture optimization, etc., and thus improve the stability, adaptability and overall operating efficiency of the satellite-to-ground collaborative network, and ensure its reliable operation in complex environments.

[0053] In an exemplary technical solution, the first satellite-to-ground network embedding integration vectors corresponding to the multiple satellite network embedding coding vectors are subjected to a second vector integration to obtain a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordinated network matrix, including: inputting the first satellite-to-ground network embedding integration vectors corresponding to the multiple satellite network embedding coding vectors into a deep residual model to generate a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordinated network matrix.

[0054] It can be understood that the core of the process of performing a second vector integration on the first satellite-to-ground network embedding integration vectors corresponding to the multiple satellite network embedding coding vectors to obtain the second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground collaborative network matrix is ​​to operate using a deep residual model.

[0055] First, we need to have a deeper understanding of the first satellite-to-ground network embedded integrated vector. The first satellite-to-ground network embedded integrated vector is obtained through a series of complex operations, and it carries the fusion information of many aspects of the state attributes of the satellite network and the ground network.

[0056] Let's take a practical example to illustrate the composition of the first satellite-to-ground network embedded integrated vector. For example, there is a specific satellite-to-ground collaborative network scenario, in which a first satellite-to-ground network embedded integrated vector contains rich dimensional information. In the dimension representing the relationship between the satellite orbit and the ground base station coverage, the value is 0.75. This means that there is a specific coverage ratio relationship between the satellite orbit and the ground base station. The value of 0.75 indicates that the satellite orbit has a high degree of coverage of the base station, covering approximately 75% of the area where the base station is located. This value has important reference value for subsequent judgments on the data transmission potential and signal coverage quality between the satellite and the ground base station.

[0057] In terms of the dimension that reflects the relationship between satellite energy reserves and ground base station power consumption, the value is 0.3. This value indicates that the satellite energy reserves are in a relatively low equilibrium state relative to the ground base station power consumption. For example, if the satellite's initial energy reserves can support its operation for 5 years, it has consumed a lot of energy and the remaining energy is relatively limited, while the power consumption of the ground base station is at a normal level. In this case, the value of 0.3 reflects the need to be cautious about energy utilization, and it may be necessary to consider optimizing the base station power consumption or adjusting the satellite's energy management strategy.

[0058] In the dimension involving the correlation between satellite sensor status and the number of user connections of ground base stations, the value is -0.4. This negative number indicates that the number of user connections of ground base stations tends to decrease based on environmental information or other related factors detected by satellite sensors. For example, satellite sensors detect that bad weather may occur in a certain area. Such weather conditions may affect the signal propagation of ground base stations, resulting in a decrease in the number of user connections. The value of -0.4 is a quantitative representation of this correlation.

[0059] In terms of the dimension that reflects the relationship between satellite communication frequency band occupancy and ground base station link status, the value is 0.65. This means that the satellite communication frequency band occupancy has a significant impact on the ground base station link status. For example, a high satellite communication frequency band occupancy rate may affect the link bandwidth, delay, and packet loss rate of the ground base station. The value of 0.65 indicates that this impact is at a medium to high level, and it is necessary to pay attention to the link status and take corresponding adjustment measures.

[0060] These are just some examples of the dimensions in the first satellite-ground network embedding integration vector. In fact, it may contain more dimensions to fully reflect the various state attribute relationships of the satellite-ground collaborative network. Under different satellite-ground collaborative network conditions, the values ​​of these dimensions will vary according to the actual network state attribute fusion results, and the value of each dimension plays a unique role in the state evaluation and subsequent operation of the entire satellite-ground collaborative network.

[0061] Next, we elaborate on how to input these first satellite-to-ground network embedding integration vectors into the deep residual model to generate the second satellite-to-ground network embedding integration vector. The deep residual model is a neural network model with a special structure, which has many advantages in processing complex input data.

[0062] The deep residual model contains multiple hidden layers, and the structure and parameter settings of these hidden layers are carefully designed according to the characteristics and processing requirements of the satellite-ground collaborative network data. When the first satellite-ground network embedding integration vector is input to the deep residual model, it first enters the input layer. For example, there are 3 first satellite-ground network embedding integration vectors, each of which has multiple dimensional features as mentioned above. These vectors are received at the input layer and are ready to be passed to the first hidden layer.

[0063] In the first hidden layer, the model extracts and transforms the input vector. The neurons in the hidden layer process the data by performing weighted summation and other operations on each feature dimension of the input. For example, for a neuron in the first hidden layer, it may assign a weight of 0.15 to the dimension that represents the relationship between the satellite orbit and the ground base station coverage in the first satellite-to-ground network embedding integrated vector, which means that when the neuron processes this dimension information, it will multiply the value of this dimension by 0.15 as part of the calculation. The dimension that reflects the relationship between satellite energy reserves and ground base station power consumption is assigned a weight of -0.08, the dimension that involves the relationship between the satellite sensor status and the number of ground base station user connections is assigned a weight of 0.25, and the dimension that reflects the relationship between satellite communication frequency band occupancy and ground base station link status is assigned a weight of 0.3. Then, the result of the weighted summation is nonlinearly transformed through a specific activation function (such as the ReLU function) to obtain the output of the neuron. The function of the ReLU function is to set values ​​less than 0 to 0 and keep values ​​greater than 0 unchanged, which can increase the nonlinear expression ability of the model.

[0064] Taking the value of the first satellite-to-ground network embedded integrated vector mentioned above as an example, for the dimension value 0.75 representing the relationship between the satellite orbit and the ground base station coverage, after being processed by the neuron, the calculation result is 0.75*0.15=0.1125, and after being processed by the ReLU function, if the result is greater than 0, it remains at 0.1125. For the dimension value 0.3 reflecting the relationship between satellite energy reserves and ground base station power consumption, the calculation result is 0.3*(-0.08)=-0.024, which becomes 0 after being processed by the ReLU function. For the dimension value -0.4 involving the association between the satellite sensor status and the number of ground base station user connections, the calculation result is -0.4*0.25=-0.1, which becomes 0 after being processed by the ReLU function. For the dimension value 0.65 reflecting the relationship between the satellite communication frequency band occupancy and the ground base station link status, the calculation result is 0.65*0.3=0.195, which is 0.195 after being processed by the ReLU function. In this way, each neuron in the first hidden layer performs a similar operation on the input first star-to-ground network embedding integration vector, thereby obtaining a set of transformed feature representations.

[0065] As data is passed through the deep residual model, the subsequent hidden layers will continue to further process the features output by the previous hidden layers. In this process, the residual connection of the deep residual model plays a crucial role. The residual connection allows information to be directly passed from the previous layer to the next layer, avoiding the problem of gradient vanishing or gradient exploding in the deep network.

[0066] For example, in the second hidden layer, in addition to receiving the output of the first hidden layer as input, some neurons also directly receive certain features of the original first satellite-to-ground network embedded integrated vector, and then combine these two parts of information for processing. For example, when a neuron in the second hidden layer processes the feature representing the relationship between the satellite orbit and the ground base station coverage, in addition to receiving the feature value 0.1125 processed by the first hidden layer, it also directly receives the feature value 0.75 in the original first satellite-to-ground network embedded integrated vector, and then combines these two values ​​according to certain rules (such as weighted average or other combination methods) to obtain a new value as the output of the neuron. This residual connection method enables the model to better retain the key features of the original input information and prevent excessive loss or distortion of information during multi-layer network transmission.

[0067] After processing through multiple hidden layers, the second satellite-ground network embedding integrated vector corresponding to the satellite-ground collaborative network matrix is ​​finally obtained at the output layer of the deep residual model. This second satellite-ground network embedding integrated vector is also a vector with multiple dimensional features, and these features are the result of deep mining and conversion of the first satellite-ground network embedding integrated vector by the deep residual model.

[0068] For example, the value of the second satellite-ground network embedding integrated vector outputted in the dimension representing the overall stability of the satellite-ground collaborative network is 0.82. This value may have changed compared to the relevant dimension value in the first satellite-ground network embedding integrated vector (if there is a similar representation). 0.82 indicates that the network as a whole is in a relatively stable state. This may be because the deep residual model comprehensively considers more factors during the processing process, such as the collaborative relationship of various state attributes between satellites and ground base stations, potential risks in the network, etc., thereby obtaining a more accurate assessment of the overall stability of the network.

[0069] The value of the dimension reflecting the satellite-to-ground data transmission efficiency is 0.55. This value indicates that the data transmission efficiency is at a medium level, which may be due to the comprehensive analysis of multiple factors in the deep residual model, such as the occupancy of satellite communication frequency bands, the link status of ground base stations, and the coverage relationship between satellites and base stations. For example, although the satellite orbit has a high degree of coverage of base stations, the occupancy of satellite communication frequency bands has certain restrictions on data transmission, resulting in the overall transmission efficiency being at a medium level.

[0070] The value of 0.72 in the dimension reflecting the degree of satellite-ground resource coordination optimization indicates a high degree of resource coordination optimization. This means that the deep residual model has determined that the overall coordination optimization of satellite-ground resources has reached a good level through the analysis of satellite energy reserves, ground base station power consumption, and other related resource utilization dimensions embedded in the first satellite-ground network integrated vector. This may be because effective resource allocation strategies have been adopted in some aspects, such as adjusting the ground base station power consumption according to the satellite energy situation, or optimizing satellite communication resource allocation according to the number of user connections required by the ground base station.

[0071] These values ​​are calculated by the deep residual model based on the input first satellite-to-ground network embedding integrated vector and its own network structure and parameters. They comprehensively reflect the overall state of the satellite-to-ground collaborative network matrix after being processed by the deep residual model. This technical solution of converting the first satellite-to-ground network embedding integrated vector into the second satellite-to-ground network embedding integrated vector through the deep residual model makes the state representation of the satellite-to-ground collaborative network more accurate and comprehensive. The deep residual model can mine deeper feature relationships in the first satellite-to-ground network embedding integrated vector, and ensure the effectiveness and stability of the model through residual connections. This provides a more reliable basis for subsequent data construction situation judgment of the satellite-to-ground collaborative network matrix based on the second satellite-to-ground network embedding integrated vector.

[0072] In the actual operation of the satellite-ground collaborative network, this precise state representation and analysis is very important. For example, when the satellite-ground collaborative network needs to optimize data transmission, based on the dimension value of data transmission efficiency in the second satellite-ground network embedded integrated vector, strategies such as satellite communication frequency band allocation and ground base station link optimization can be adjusted in a targeted manner. If the overall stability of the network is to be improved, based on the relevant values ​​in the second satellite-ground network embedded integrated vector, satellite orbit adjustment and ground base station layout optimization can be used.

[0073] During the data construction process of the satellite-ground collaborative network, the second satellite-ground network embedded integrated vector is obtained by performing a second vector integration on the first satellite-ground network embedded integrated vector through a deep residual model, which can more accurately reflect the status of the satellite-ground collaborative network and provide strong support for the management, optimization and stable operation of the entire satellite-ground collaborative network.

[0074] It can be seen that, firstly, the deep residual model is used to process the first satellite-ground network embedding integrated vector to obtain the second satellite-ground network embedding integrated vector. The structural characteristics of the deep residual model enable it to effectively mine the deep-level feature relationship of the data. After being processed, the multi-dimensional features in the first satellite-ground network embedding integrated vector are used in the second satellite-ground network embedding integrated vector to reflect the different aspects of the satellite-ground collaborative network with new values, such as stability, transmission efficiency, and resource collaborative optimization degree. This helps to grasp the overall state of the satellite-ground collaborative network more accurately and provide a more reliable basis for situation judgment. Secondly, the residual connection mechanism in the deep residual model avoids the common gradient vanishing or gradient explosion problems in deep networks, ensures the effectiveness and stability of the model, thereby improving the reliability and accuracy of the entire satellite-ground collaborative network data construction process, and helps to optimize the management and operation of the satellite-ground collaborative network, so that it can better adapt to the complex and changing network environment.

[0075] In another exemplary technical scheme, before the first satellite-to-ground network embedding integration vectors corresponding to the multiple satellite network embedding coding vectors are subjected to a second vector integration to obtain the second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordination network matrix, it also includes: obtaining at least one group of first satellite side status data corresponding to a target multi-task information observation and scheduling server from the multiple groups of satellite side status data, wherein the target multi-task information observation and scheduling server refers to a multi-task information observation and scheduling server that has a spatiotemporal involvement relationship with the satellite-to-ground coordination network matrix; determining a first satellite-to-ground coordination confidence coefficient corresponding to the at least one group of first satellite side status data; determining a second satellite-to-ground coordination confidence coefficient corresponding to the second satellite side status data, wherein the second satellite side status data refers to satellite side status data in the multiple groups of satellite side status data except the at least one group of first satellite side status data, and the first satellite-to-ground coordination confidence coefficient is different from the second satellite-to-ground coordination confidence coefficient.

[0076] Based on this, the method of performing a second vector integration on the first satellite-to-ground network embedded integration vectors corresponding to the multiple satellite network embedded coding vectors respectively to obtain a second satellite-to-ground network embedded integration vector corresponding to the satellite-to-ground coordinated network matrix includes: performing the first vector integration on the satellite network embedded coding vector corresponding to the at least one group of first satellite side status data and the ground network embedded coding vector according to the first satellite-to-ground coordination confidence coefficient to obtain a first satellite-to-ground network embedded integration vector corresponding to the at least one group of first satellite side status data; performing the first vector integration on the satellite network embedded coding vector corresponding to the second satellite side status data and the ground network embedded coding vector according to the second satellite-to-ground coordination confidence coefficient to obtain a first satellite-to-ground network embedded integration vector corresponding to the second satellite side status data.

[0077] In the next step, determining the first satellite-ground coordination confidence coefficient corresponding to the at least one group of first satellite side status data includes: determining the number of groups of the at least one group of first satellite side status data; determining the probability heat map frequent item features and the probability heat map edge item features based on the number of groups; configuring the first satellite-ground coordination confidence coefficient for the at least one group of first satellite side status data according to the timing priority interval of the at least one group of first satellite side status data using probability thermodynamic statistical rules, wherein the first satellite side status data located within the timing priority interval is configured with the probability heat map frequent item features, and the first satellite side status data located at the boundary of the timing priority interval is configured with the probability heat map edge item features.

[0078] Under this exemplary technical solution, the overall technical solution revolves around a series of operations before the second vector integration and a specific method of the second vector integration. These operations and methods are critical to the data construction of the satellite-ground collaborative network matrix.

[0079] First, the operation of obtaining at least one set of first satellite-side status data corresponding to the target multi-task information and observation scheduling server is described. In the complex architecture of the satellite-ground collaborative network, the multi-task information and observation scheduling server plays a key role, responsible for coordinating and managing various tasks and resources in the network. The target multi-task information and observation scheduling server refers to a multi-task information and observation scheduling server that has a spatiotemporal relationship with the satellite-ground collaborative network matrix. The spatiotemporal relationship here is a complex concept involving multiple factors.

[0080] From a spatial perspective, in the satellite-ground collaborative network, the vast area covered by the satellite network is intertwined with the distribution area of ​​the ground network. Different multi-task information and observation scheduling servers may be responsible for different spatial areas. For example, a multi-task information and observation scheduling server may be mainly responsible for an area within a certain longitude and latitude range, which may include multiple ground base stations and satellite coverage. When the satellite network and the ground network in this area are closely connected in terms of data interaction, task scheduling, etc., the multi-task information and observation scheduling server has a spatial involvement relationship with the satellite-ground collaborative network matrix.

[0081] From a temporal perspective, the execution of tasks has temporal sequence and periodic characteristics. In the same time period, different tasks may require different resource scheduling and coordination. For example, in a certain time period, there may be a large number of data transmission tasks from satellites to ground base stations. At this time, the multi-task information and observation scheduling server responsible for scheduling these tasks has a temporal involvement relationship with the satellite-ground collaborative network matrix. When a multi-task information and observation scheduling server has such a close involvement relationship with the satellite-ground collaborative network matrix in both space and time, it is determined as the target multi-task information and observation scheduling server.

[0082] At least one set of first satellite side status data is obtained from the multiple sets of satellite side status data corresponding to the target multi-task information observation scheduling server. This set of data contains rich satellite status information associated with the target multi-task information observation scheduling server. These satellite status information can cover multiple aspects, such as the orbital parameters of the satellite, the operating status of each device on the satellite, the communication link status of the satellite, etc. Taking the orbital parameters of the satellite as an example, it may include information such as the orbital altitude, orbital inclination, and orbital period of the satellite. For example, the orbital altitude of a satellite is 800 kilometers, the orbital inclination is 50 degrees, and the orbital period is 100 minutes. These values ​​are all part of the satellite side status data and are closely related to the scheduling tasks of the target multi-task information observation scheduling server. For example, the orbital altitude affects the communication distance and signal propagation delay between the satellite and the ground base station, the orbital inclination may affect the coverage range and coverage time of the satellite to different areas on the ground, and the orbital period is related to the satellite's repeated visit cycle and data collection frequency.

[0083] Then, a first satellite-ground coordination confidence coefficient corresponding to at least one set of first satellite-side status data and a second satellite-ground coordination confidence coefficient corresponding to the second satellite-side status data are determined (the second satellite-side status data refers to the satellite-side status data in multiple sets of satellite-side status data except for at least one set of first satellite-side status data, and the first satellite-ground coordination confidence coefficient is different from the second satellite-ground coordination confidence coefficient).

[0084] When determining the first satellite-ground coordination confidence coefficient corresponding to at least one set of first satellite-side status data, multiple steps are included. First, the number of groups of at least one set of first satellite-side status data is determined. This operation is based on the classification and statistics of the satellite-side status data. For example, after careful analysis and statistics, it is determined that the number of groups of at least one set of first satellite-side status data is 5 groups. These 5 groups of data are processed differently in subsequent operations, and the determination of the number of groups provides basic information for the subsequent determination of the confidence coefficient.

[0085] Then, the probability heat map frequent item features and the probability heat map edge item features are determined according to the number of groups. In this process, the probability heat map is a tool for describing data distribution and association relationships. According to the pre-set rules, when the number of groups is 5, the probability heat map frequent item features may be determined as a value indicating a higher degree of correlation or importance, such as 0.75. This value indicates that in the satellite-ground collaborative network, this part of the first satellite side status data associated with the target multi-task information observation scheduling server has a higher correlation or importance in certain key factors. For example, these data may be closely related to the current main task (such as a high-priority data transmission task), or play an important role in maintaining the key functions of the satellite-ground collaborative network (such as communication coverage in a specific area).

[0086] The edge feature of the probability heat map may be a value indicating relatively low relevance or importance, such as 0.15. This means that in the operation and data construction process of the entire satellite-ground collaborative network, although these data also have a certain role, they are relatively less important or relevant. For example, these data may be some auxiliary monitoring data, or data that will only play a role in special circumstances.

[0087] Then, according to the timing priority interval of at least one set of first satellite side status data, the first satellite-ground coordination confidence coefficient is configured for at least one set of first satellite side status data using the probabilistic thermodynamic statistical rule. The timing priority interval is a time range set according to the task requirements and data characteristics of the satellite-ground coordination network. For example, the timing priority interval is set to [0, 20] (the numerical value here represents a certain time unit, such as minutes). The first satellite side status data within this timing priority interval is configured with the probabilistic heat map frequent item feature, i.e., 0.75.

[0088] For example, for a set of satellite-side status data with a timestamp of 10 minutes, this set of data is configured with a first satellite-ground coordination confidence coefficient of 0.75. This is because the satellite-side status data within this time range may be closely related to the critical tasks currently being performed or important network status. The first satellite-side status data located at the boundary of the timing priority interval is configured with the probability heat map edge item feature, that is, 0.15. For example, the satellite-side status data with a timestamp of 0 or 20 minutes is configured with a first satellite-ground coordination confidence coefficient of 0.15. This way of configuring confidence coefficients based on the timing priority interval can better take into account the dynamic characteristics of the satellite-side status data in time and its relevance to the satellite-ground coordination network tasks.

[0089] Based on the confidence coefficient determined above, a first vector integration operation is performed. According to the first satellite-ground coordination confidence coefficient, a satellite network embedding coding vector and a ground network embedding coding vector corresponding to at least one set of first satellite side state data are integrated by a first vector to obtain a first satellite-ground network embedding integration vector corresponding to at least one set of first satellite side state data.

[0090] Take an example to illustrate, for example, a dimension in the satellite network embedded coding vector represents the energy reserve of the satellite, and the value is 80% (indicating that the satellite still has 80% of its energy remaining), and a dimension in the ground network embedded coding vector represents the available bandwidth of the ground base station, and the value is 50Mbps. When the first satellite-ground coordination confidence coefficient is 0.75, when the first vector is integrated, the weight of the dimension of satellite energy reserve in the integrated vector will be relatively high. Specifically, it may participate in the construction of the first satellite-ground network embedded integrated vector according to a certain calculation method of 0.75*80=60; while the weight of the dimension of the available bandwidth of the ground base station is relatively low, and participates in the construction according to a certain calculation method of 0.25*50=12.5 (for example, the sum of the overall weight is 1, and this is just an example of the weight distribution relationship). In this way, the first satellite-ground network embedded integrated vector corresponding to at least one set of first satellite side status data is obtained, which integrates the information of the satellite network and the ground network, and adjusts the weights of the information of different dimensions according to the confidence coefficient.

[0091] Similarly, the satellite network embedding code vector and the ground network embedding code vector corresponding to the second satellite side state data are integrated by the first vector according to the second satellite-ground coordination confidence coefficient to obtain the first satellite-ground network embedding integration vector corresponding to the second satellite side state data. For example, the second satellite-ground coordination confidence coefficient is 0.3, for the satellite energy reserve dimension in the satellite network embedding code vector (the value is still 80%) and the available bandwidth dimension of the ground base station in the ground network embedding code vector (the value is still 50Mbps), when performing the first vector integration, the dimension of the satellite energy reserve is involved in the construction of the first satellite-ground network embedding integration vector according to a certain calculation method of 0.3*80=24, and the dimension of the available bandwidth of the ground base station is involved in the construction according to a certain calculation method of 0.7*50=35 (same as the sum of the overall weights is 1).

[0092] After completing the above-mentioned first vector integration for different satellite-side status data, different first satellite-to-ground network embedding integration vectors are obtained, and then these vectors will be used for the second vector integration to obtain the second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground collaborative network matrix. In this way, satellite-side status data under different confidence coefficients are treated differently in the first vector integration, which in turn affects the generation of the final second satellite-to-ground network embedding integration vector. This operation based on different confidence coefficients can more accurately reflect the relationship between different satellite status data and the ground network in the satellite-to-ground collaborative network, thereby improving the accuracy and effectiveness of the entire satellite-to-ground collaborative network matrix data construction.

[0093] In the actual operation of the satellite-ground collaborative network, different satellite status data have different impacts on the entire network due to their different relationships with the target multi-task information and observation scheduling server and differences in timing. For example, during an urgent data transmission task, satellite status data that is closely related to the target multi-task information and observation scheduling server and is in the timing priority interval (such as the satellite's communication link is in good condition and has sufficient energy) is crucial to ensure the smooth progress of the task. These data are given a larger weight in the vector integration through a higher confidence coefficient, so that their importance can be better reflected when constructing the first satellite-ground network embedded integration vector and the final second satellite-ground network embedded integration vector. Although some relatively minor satellite status data (such as the status data of some auxiliary monitoring equipment) are also involved in the entire data construction process, due to their lower confidence coefficients, their weight in the vector integration is relatively small, and will not cause excessive interference to key tasks and the main functions of the network.

[0094] In this way, by determining different satellite-ground coordination confidence coefficients before the first vector integration, the satellite side status data is processed differently, and then the first vector integration and the subsequent second vector integration are performed, so that the final second satellite-ground network embedded integration vector can better reflect the complex relationship of the satellite-ground coordination network. It helps to make more reasonable use of these satellite status data and improve the quality of data construction in the process of satellite-ground coordination network data construction, and provide a more reliable basis for subsequent operations based on the second satellite-ground network embedded integration vector (such as data construction situation judgment, etc.). At the same time, this method also helps to optimize the resource allocation of the satellite-ground coordination network, improve the overall performance and stability of the network, and enable it to better adapt to complex and changing network environments and mission requirements.

[0095] In this way, first of all, by determining the confidence coefficient of the satellite-side status data related to the target multi-task information observation scheduling server, the importance of different satellite status data to the satellite-ground collaborative network can be distinguished. For example, different confidence coefficients are determined according to the timing priority interval and the number of groups, so that data processing is more targeted and can better adapt to the complex task requirements and dynamic changes of data in the satellite-ground collaborative network. Secondly, the first vector integration based on different confidence coefficients can reasonably adjust the weights of satellite network and ground network information in the integrated vector, and more accurately reflect the relationship between the two, thereby improving the accuracy of the first satellite-ground network embedded integrated vector. This helps to obtain the second satellite-ground network embedded integrated vector that can better reflect the actual situation of the satellite-ground collaborative network in the subsequent second vector integration. Finally, this method of accurately constructing the second satellite-ground network embedded integrated vector provides a more reliable basis for the construction of the satellite-ground collaborative network matrix data, helps to optimize the management and operation of the network, and improves the overall performance, stability and adaptability of the network to complex tasks and environmental changes.

[0096] In some preferred embodiments, the satellite-ground collaborative network matrix includes multiple network matrix blocks, wherein the x-th network matrix block corresponds to the x-th group of satellite side status data, and x is a positive integer; before mining the satellite network embedding coding vectors corresponding to the multiple groups of satellite side status data, it also includes: determining the y-th group of initial satellite side status data in the x-th network matrix block as the x-th group of satellite side status data, and y is a positive integer.

[0097] In the actual application process, the structural characteristics of the satellite-ground collaborative network matrix become an important basis for the entire technical solution. The satellite-ground collaborative network matrix contains multiple network matrix blocks. This block structure is designed to more finely process and analyze the data and state relationship in the satellite-ground collaborative network.

[0098] Each network matrix block corresponds to a set of satellite-side status data. The corresponding relationship here is not set arbitrarily, but is based on the operating mechanism and data association logic of the satellite-ground collaborative network. For example, the first network matrix block corresponds to the first set of satellite-side status data, the second network matrix block corresponds to the second set of satellite-side status data, and so on, where x is a positive integer. This corresponding relationship runs through the data processing flow of the entire technical solution.

[0099] Before mining the satellite network embedding code vectors corresponding to multiple groups of satellite side state data, there is an operation of determining the yth group of initial satellite side state data in the xth network matrix block as the xth group of satellite side state data, where y is a positive integer. The significance of this operation is to select the data group closely related to the specific network matrix block from the numerous initial satellite side state data, so as to conduct more accurate satellite network embedding code vector mining in the future.

[0100] Take a specific satellite-ground collaborative network matrix structure as an example. For example, this satellite-ground collaborative network matrix is ​​divided into 5 network matrix blocks. For the third network matrix block, it contains multiple sets of initial satellite-side status data. These initial satellite-side status data cover multiple aspects of satellite information. For example, one set of initial satellite-side status data may include the orbital position information of the satellite, with an orbital altitude of 700 kilometers and an orbital inclination of 45 degrees; in terms of the satellite's communication status, the communication frequency band is the X band and the signal strength is -80dBm; in terms of the satellite's energy status, the energy reserve is 60%, etc. In this network matrix block, after specific rules and logic, one of the sets of initial satellite-side status data is determined as the third set of satellite-side status data corresponding to this network matrix block.

[0101] This determination process is based on many factors. From the perspective of the satellite's orbital position, satellites in certain orbital positions may have a more direct relationship with the ground area or communication link involved in a specific network matrix block. For example, a satellite with an orbital altitude of 700 kilometers and an orbital inclination of 45 degrees may have a unique geometric relationship when covering the ground area corresponding to a specific network matrix block, making its status data more critical to the network matrix block. From the perspective of communication status, a satellite with X-band communication and a signal strength of -80dBm may play an important role in the communication tasks related to the network matrix block, and its communication status data is indispensable for understanding and processing the network characteristics of this network matrix block. Similarly, the status of a satellite with an energy reserve of 60% may also be closely related to the task scheduling and resource allocation in the network matrix block. For example, some tasks may require the satellite to have a certain amount of energy reserves to execute, and the 60% energy reserve may just meet the requirements of some tasks in the network matrix block.

[0102] After the xth set of satellite state data is determined, the satellite network embedded code vector can be mined. This mining process converts the satellite state data into a form that can better process and analyze data in the satellite-ground collaborative network. For each set of satellite state data, the mined satellite network embedded code vector contains the abstraction and encoding of the satellite state data features.

[0103] Taking the third group of satellite-side status data mentioned above as an example, when mining the satellite network embedded coding vector, the information that the satellite's orbital altitude is 700 kilometers may be encoded as a specific value or coding segment. For example, in the coding system, the orbital altitude between 500-1000 kilometers is encoded as a specific coding segment, which may correspond to a number or a set of binary codes, such as 101. The orbital inclination of 45 degrees may be encoded as another specific coding form, such as 011. For the communication frequency band of the X band, it may be encoded as 110, and the signal strength of -80dBm may be encoded as a code divided according to the signal strength range, such as 001. The satellite energy reserve of 60% will also be encoded as a code corresponding to the energy reserve percentage, such as 100. These codes are combined to form the satellite network embedded coding vector corresponding to the third group of satellite-side status data.

[0104] In this way, the satellite-side status data corresponding to each network matrix block is converted into a satellite network embedded coding vector, which can more effectively perform subsequent data processing operations. For example, when merging or integrating with the ground network embedded coding vector, these coding vectors can participate in the operation in a unified form that is easy to calculate and analyze. Moreover, this coding method can better reflect the inherent connection between the satellite-side status data and each block of the satellite-ground collaborative network matrix, providing a more accurate and effective data foundation for the data construction of the entire satellite-ground collaborative network.

[0105] This technical solution of associating network matrix blocks with satellite-side status data, determining satellite-side status data in a specific way, and then mining the corresponding satellite network embedded coding vectors can help improve the accuracy and effectiveness of satellite-ground collaborative network data processing. In the actual operation of the satellite-ground collaborative network, this precise data processing can better adapt to complex network environments and diverse mission requirements. For example, when allocating resources, satellite resources and ground network resources can be allocated more reasonably based on accurate satellite network embedded coding vectors; in terms of network optimization, satellites and ground networks related to specific network matrix blocks can be optimized and adjusted more specifically, thereby improving the performance and stability of the entire satellite-ground collaborative network.

[0106] In a replaceable technical idea, the data building situation of the satellite-ground collaborative network matrix is ​​judged based on the second satellite-ground network embedded integrated vector to obtain the building situation decision information corresponding to the satellite-ground collaborative network matrix, including: inputting the second satellite-ground network embedded integrated vector into a pre-debugged data building situation judgment network to generate the building situation decision information corresponding to the satellite-ground collaborative network matrix.

[0107] In addition, before embedding the second satellite-to-ground network into the integrated vector input data to build a situation discrimination network and generating the building situation decision information corresponding to the satellite-to-ground collaborative network matrix, it also includes: obtaining a ground network equipment monitoring data example and a plurality of groups of satellite side status data examples corresponding to the satellite-to-ground collaborative network matrix example, wherein the ground network equipment monitoring data example is the ground network equipment monitoring data involved in the data building task process of the satellite-to-ground collaborative network matrix example, and the satellite-to-ground collaborative network matrix example carries a building situation decision annotation; mining the ground network embedding coding vector corresponding to the ground network equipment monitoring data example, and mining the satellite network embedding coding vectors corresponding to the plurality of groups of satellite side status data examples respectively; embedding a plurality of satellites into the integrated vector input data; and generating the building situation decision information corresponding to the satellite-to-ground collaborative network matrix example. The satellite network embedded coding vector and the ground network embedded coding vector are respectively integrated with the first vector to obtain the first satellite-to-ground network embedded integration vectors corresponding to the multiple groups of satellite side status data examples; the first satellite-to-ground network embedded integration vectors corresponding to the multiple groups of satellite side status data examples are integrated with the second vector to obtain the second satellite-to-ground network embedded integration vector corresponding to the satellite-to-ground collaborative network matrix example; the second satellite-to-ground network embedded integration vector is input into the initial situation discrimination network to generate the construction situation prediction information corresponding to the satellite-to-ground collaborative network matrix example; the initial situation discrimination network is debugged based on the difference between the construction situation prediction information and the construction situation decision annotation to obtain the data construction situation discrimination network.

[0108] In this replaceable technical idea, the entire technical solution revolves around judging the data construction situation of the satellite-ground collaborative network matrix based on the second satellite-ground network embedded integrated vector to obtain construction situation decision information, which includes multiple key operation steps and network debugging process.

[0109] First, the core operation of obtaining the construction situation decision information by judging the data construction situation of the satellite-ground collaborative network matrix based on the second satellite-ground network embedding integrated vector is described, that is, the second satellite-ground network embedding integrated vector is input into the pre-debugged data construction situation judgment network, thereby generating the construction situation decision information corresponding to the satellite-ground collaborative network matrix. This data construction situation judgment network is obtained through a series of pre-debugging processes, and is specifically used to accurately judge the data construction situation of the satellite-ground collaborative network matrix based on the input second satellite-ground network embedding integrated vector and output the corresponding decision information.

[0110] Before the second satellite-to-ground network is embedded in the integrated vector input data to build the situation discrimination network, there are a series of operations to build and debug this discrimination network. The first is to obtain the ground network equipment monitoring data examples and multiple sets of satellite side status data examples corresponding to the satellite-to-ground collaborative network matrix example. The satellite-to-ground collaborative network matrix example here is a representative satellite-to-ground collaborative network structure example, and the ground network equipment monitoring data involved in the data building task process is defined as a ground network equipment monitoring data example. These ground network equipment monitoring data contain information from multiple aspects, such as the equipment operation status data of the ground base station. Taking a ground base station as an example, the temperature data in its equipment operation status data may be 30 degrees Celsius, the power consumption data is 2 kilowatts, and the number of user connections within the signal coverage range is 500. At the same time, there are multiple sets of satellite side status data examples, which contain multiple status information of the satellite. For example, the orbital altitude of a satellite is 600 kilometers, the sampling frequency of a sensor on the satellite is 10 times per second, and the energy reserve of the satellite is 70%. In addition, the satellite-ground collaborative network matrix example carries a construction situation decision annotation, which is a known and accurate judgment result of the construction situation of the satellite-ground collaborative network matrix data of this example, and is used for subsequent debugging of the initial situation judgment network.

[0111] The next step is to mine the ground network embedded coding vector corresponding to the ground network equipment monitoring data example, and to mine the satellite network embedded coding vector corresponding to multiple groups of satellite side status data examples. For the ground base station data in the ground network equipment monitoring data example, different data will be encoded when mining the ground network embedded coding vector. For example, for the temperature data of 30 degrees Celsius, it may be encoded as a specific value according to the pre-set temperature interval coding rule, such as the temperature between 25-35 degrees Celsius is encoded as 10; for the power consumption data of 2 kilowatts, according to the power consumption interval coding rule, for example, 1-3 kilowatts is encoded as 01; for the number of user connections of 500, according to the user connection number interval coding rule, for example, 300-600 is encoded as 11. These codes are combined to form the ground network embedded coding vector. The satellite data in the satellite side status data example is also encoded. For example, the satellite orbit height is 600 km, according to the orbit height interval coding rule, for example, 500-700 km is coded as 010; the satellite sensor sampling frequency is 10 times per second, according to the sampling frequency interval coding rule, for example, 5-15 times per second is coded as 101; the satellite energy reserve is 70%, according to the energy reserve percentage interval coding rule, for example, 60%-80% is coded as 110. These codes constitute the satellite network embedded coding vector.

[0112] Then, the first vector integration is performed on the multiple satellite network embedded coding vectors and the ground network embedded coding vectors respectively, and the first satellite-ground network embedded integration vectors corresponding to the multiple groups of satellite side status data examples are obtained. Taking two groups of satellite network embedded coding vectors and ground network embedded coding vectors as examples, when performing the first vector integration, the elements in each vector are combined and calculated according to certain integration rules. For example, for the elements in the satellite network embedded coding vector of satellite 1 and the elements in the ground network embedded coding vector, summation or other calculation operations may be performed according to a certain weight distribution. For example, the satellite network embedding coding vector of satellite 1 is [010, 101, 110], and the ground network embedding coding vector is [10, 01, 11]. When the first vector is integrated, the elements at each corresponding position are calculated according to specific weights. For example, the weight of the element at the first position is 0.3, the weight of the element at the second position is 0.3, and the weight of the element at the third position is 0.4. Then, after calculation, the first element in the first satellite-to-ground network embedding integrated vector corresponding to satellite 1 is 0.3*010+0.1*10=001 (the actual calculation can be more complicated according to the specific coding and weight definition), and so on to obtain the complete first satellite-to-ground network embedding integrated vector.

[0113] Then, the first satellite-to-ground network embedding integrated vectors corresponding to the multiple groups of satellite-side status data examples are respectively integrated by the second vector to obtain the second satellite-to-ground network embedding integrated vector corresponding to the satellite-to-ground coordinated network matrix example. Similarly, taking two groups of first satellite-to-ground network embedding integrated vectors as an example, for example, the first satellite-to-ground network embedding integrated vector 1 is [001, 110, 011], and the first satellite-to-ground network embedding integrated vector 2 is [101, 010, 100]. When performing the second vector integration, a rule different from the first vector integration will be adopted. For example, some elements may be enlarged or reduced first, and then summed or other complex combination operations are performed. For example, the first element 001 in the first satellite-to-ground network embedding integrated vector 1 is multiplied by 2 to obtain 010, and the first element 101 in the first satellite-to-ground network embedding integrated vector 2 is divided by 2 to obtain 010, and then the two adjusted elements are added to obtain 010+010=100 (this is just an exemplary introduction, and the actual operation is more complicated), and other elements are operated by analogy, and finally the second satellite-to-ground network embedding integrated vector corresponding to the satellite-to-ground coordinated network matrix example is obtained.

[0114] Then, the second satellite-ground network embedding integrated vector is input into the initial situation discrimination network to generate the construction situation prediction information corresponding to the satellite-ground collaborative network matrix example. This initial situation discrimination network is a preliminary constructed network for judging the situation, and its structure and parameters may be based on some general design principles or experience settings. When the second satellite-ground network embedding integrated vector is input, the network will output a construction situation prediction information according to its internal calculation logic. This prediction information is a preliminary judgment result of the data construction situation of the satellite-ground collaborative network matrix example.

[0115] Finally, the initial situation discrimination network is debugged based on the difference between the construction situation prediction information and the construction situation decision annotation to obtain the data construction situation discrimination network. Due to the preliminary nature of the initial situation discrimination network, the construction situation prediction information it outputs may differ from the known accurate construction situation decision annotation. For example, the construction situation prediction information may judge that the satellite-ground collaborative network matrix example is in a stable data construction situation, while the construction situation decision annotation indicates that it is in a data construction situation that needs to be optimized. By analyzing this difference, the structural parameters of the initial situation discrimination network (such as the connection weights of neurons, the parameters of the activation function, etc.) are adjusted. After multiple such adjustment processes, a large number of satellite-ground collaborative network matrix examples and their corresponding ground network equipment monitoring data examples, satellite-side status data examples, construction situation decision annotations, etc. are used to continuously optimize the initial situation discrimination network until the difference between its output construction situation prediction information and construction situation decision annotations reaches an acceptable range, thereby obtaining a pre-debugged data construction situation discrimination network, which can more accurately judge the data construction situation of the satellite-ground collaborative network matrix based on the input second satellite-ground network embedding integrated vector and generate corresponding construction situation decision information.

[0116] With this design, first of all, by using the satellite-ground collaborative network matrix example and its related data to build and debug the data construction situation discrimination network, the accuracy of network judgment can be improved. For example, the integrated vector obtained by mining the encoding vector and performing vector integration based on the data in the example is used to debug the network. Secondly, the network is adjusted according to the difference between the construction situation prediction information and the decision annotation, which can continuously optimize the network performance. Finally, the pre-debugged network can accurately generate the construction situation decision information of the satellite-ground collaborative network matrix based on the second satellite-ground network embedded integrated vector, which helps to effectively manage the construction of the satellite-ground collaborative network data.

[0117] In some independent embodiments, after performing data construction situation judgment on the satellite-ground collaborative network matrix based on the second satellite-ground network embedded integrated vector to obtain construction situation decision information corresponding to the satellite-ground collaborative network matrix, the method also includes: based on the construction situation decision information, constructing a data relationship network for the ground-side perception data and the multiple groups of satellite-side status data to generate a full-network perception data relationship network corresponding to the satellite-ground collaborative network matrix.

[0118] In detail, based on the situation decision information, a data relationship network is constructed for the ground-side perception data and the multiple groups of satellite-side status data to generate a full-network perception data relationship network corresponding to the satellite-ground collaborative network matrix, specifically including: using the situation decision information to extract the ground network perception data topology and the star network status data topology of the satellite-ground collaborative network matrix, the ground network perception data topology is used to characterize the ground-side perception data corresponding to the satellite-ground collaborative network matrix, and the star network status data topology is used to characterize the multiple groups of satellite-side status data corresponding to the satellite-ground collaborative network matrix; for the ground network perception data topology, based on the data topology associated with the ground network perception data topology in the star network status data topology Node set, obtain a first full-network perception linkage node set, the first full-network perception linkage node set is used to represent the ground network perception data topology that integrates the star network status data topology; for the star network status data topology, based on the data topology node set associated with the star network status data topology in the ground network perception data topology, obtain a second full-network perception linkage node set, the second full-network perception linkage node set is used to represent the star network status data topology that integrates the ground network perception data topology; establish a relationship between the first full-network perception linkage node set and the second full-network perception linkage node set to obtain a full-network perception linkage relationship topology; based on the full-network perception linkage relationship topology, generate a full-network perception data relationship network corresponding to the star-ground collaborative network matrix.

[0119] In the above-mentioned independent embodiment, after the entire technical scheme performs data construction situation judgment on the satellite-ground collaborative network matrix according to the second satellite-ground network embedded integrated vector to obtain construction situation decision information, it further performs a series of operations based on the construction situation decision information to generate a full-network perception data relationship network corresponding to the satellite-ground collaborative network matrix.

[0120] First, after obtaining the construction situation decision information, the ground network perception data topology and star network status data topology of the satellite-ground collaborative network matrix are extracted based on the information. The ground network perception data topology is used to characterize the ground-side perception data corresponding to the satellite-ground collaborative network matrix, while the star network status data topology is used to characterize multiple groups of satellite-side status data corresponding to the satellite-ground collaborative network matrix.

[0121] Taking a specific satellite-ground collaborative network matrix as an example, the ground-side perception data contains multiple aspects of information, which reflects the operating status of ground network equipment and the relationship between it and other equipment. For example, consider a ground base station group, in which the temperature of a ground base station is 25 degrees Celsius. This temperature data is an important part of many ground-side perception data. In terms of power consumption, the power consumption of this base station is 3 kilowatts, which reflects the energy consumption of the base station during operation. Looking at the number of user connections within the signal coverage area, it is 800. This data reflects the load of the base station and its service capabilities in the area. There is an intrinsic connection between these data. When constructing the ground network perception data topology, it will be constructed based on these data and the relationship between them.

[0122] When constructing the ground network perception data topology, the temperature, power consumption, number of user connections, etc. may be used as nodes, and the relationship between them may be used as edges to construct the topology. For example, temperature may affect the performance of base station equipment, and thus affect power consumption, so in the topology, there is an edge between the temperature node and the power consumption node to indicate this influence relationship between them. Similarly, changes in the number of user connections may cause fluctuations in power consumption, so there is also an edge between the user connection number node and the power consumption node. This topology can clearly show the relationship between ground-side perception data and provide a basis for subsequent operations.

[0123] For the star network status data topology, the satellite side status data also contains rich and diverse information. For example, the orbital altitude of a certain satellite is 700 kilometers. This orbital altitude data has an important impact on the satellite's coverage, communication delay, etc. The sampling frequency of a certain sensor on the satellite is 15 times per second. This sampling frequency reflects the satellite's ability and frequency to collect external environment or other data. The satellite's energy reserve is 65%. This data reflects the satellite's energy status and plays a key role in the satellite's continued operation and function execution.

[0124] When constructing the star network status data topology, the satellite's orbital height, sensor sampling frequency, energy reserves, etc. are used as nodes, and the relationship between them is used as edges to construct the topological structure. For example, the satellite's orbital height may affect the sampling frequency of the sensor, because different orbital heights correspond to different observation angles and environmental conditions, which affect the working efficiency of the sensor. Therefore, in the star network status data topology, there is an edge between the orbital height node and the sampling frequency node to represent this influence relationship. The satellite's energy reserves are also related to the work of the sensor. The amount of energy reserves will limit the working time and sampling frequency of the sensor, so there is also an edge between the energy reserve node and the sampling frequency node.

[0125] Next, for the ground network perception data topology, based on the data topology node set associated with the ground network perception data topology in the star network status data topology, the first full network perception linkage node set is obtained. This operation aims to integrate the relevant information in the star network status data topology into the ground network perception data topology, thereby constructing a node set that more comprehensively reflects the satellite-ground collaborative network relationship.

[0126] For example, the satellite orbit height node in the star network status data topology may be associated with the ground base station coverage range node in the ground network perception data topology. When the satellite orbit height is 700 kilometers, the ground area it covers is specific, and this coverage area overlaps or affects the coverage range of the ground base station. Because the coverage range of the satellite will affect the satellite signal strength and quality that the ground base station can receive, which in turn affects the number of user connections and other related data of the ground base station. These related nodes are integrated together to form the first full-network perception linkage node set. This node set not only includes the original nodes in the ground network perception data topology, but also incorporates the nodes in the star network status data topology related to the ground network perception data topology. It represents the ground network perception data topology that integrates the star network status data topology.

[0127] In this process, it is necessary to carefully analyze the relationship between each node in the star network status data topology and the ground network perception data topology. For example, although the satellite's energy reserve node does not seem to have a direct relationship with the ground base station's temperature node, through the satellite's communication function (satellite energy reserves affect the communication duration and frequency, which in turn affects the stability of the ground base station receiving satellite signals, and the stability of the ground base station receiving signals may affect the heat dissipation of the base station equipment, thereby indirectly affecting the temperature), an indirect relationship between them can be found. These indirectly related nodes are also reasonably integrated into the first full-network perception linkage node set, so that this node set can more comprehensively reflect the complex relationship in the satellite-ground collaborative network.

[0128] Similarly, for the star network status data topology, based on the data topology node set associated with the star network status data topology in the ground network perception data topology, a second full network perception linkage node set is obtained. For example, the ground base station power consumption node in the ground network perception data topology may be associated with the satellite energy reserve node in the star network status data topology. The power consumption of the ground base station is 3 kilowatts, and this power consumption requirement may affect the satellite's energy allocation strategy. If the power consumption demand of the ground base station is high, the satellite may need to adjust its communication power or data transmission strategy, which will affect the satellite's energy consumption. Therefore, the satellite's energy reserve needs to be reasonably planned and adjusted according to the power consumption of the ground base station.

[0129] These related nodes are integrated to obtain the second full-network perception linkage node set, which represents the star network status data topology that integrates the ground network perception data topology. When constructing this node set, various indirect relationships also need to be considered. For example, the number of user connections to the ground base station may affect the energy reserves of the satellite by affecting the power consumption of the ground base station, and then affect the sampling frequency of the satellite sensor through the satellite energy reserves. Therefore, in the second full-network perception linkage node set, these related nodes need to be included to fully reflect this complex relationship chain.

[0130] Then, the relationship between the first whole-network perception linkage node set and the second whole-network perception linkage node set is constructed to obtain the whole-network perception linkage relationship topology. This process is to further sort out and construct the relationship between the two node sets that have integrated some information, forming a more complete and accurate topological structure that reflects the various data relationships in the satellite-ground collaborative network.

[0131] For example, there may be an indirect relationship between the ground base station temperature node in the first full-network perception linkage node set and the satellite sensor sampling frequency node in the second full-network perception linkage node set. When the ground base station temperature is 25 degrees Celsius, the performance of the base station equipment is in a state that affects the signal transmission quality and efficiency of the base station. The satellite sensor sampling frequency is 15 times per second, and the information collected by the satellite sensor is transmitted to the ground base station through the satellite communication link. If the signal transmission quality and efficiency of the ground base station change, it may affect the transmission and processing of the satellite sensor collected information, which may cause the satellite to adjust the sampling frequency of the sensor. By analyzing their association with other nodes (such as the ground base station temperature may affect the number of user connections, the number of user connections is related to the satellite communication task volume, and the satellite communication task volume affects the sampling frequency of the satellite sensor), the relationship edges between them are constructed in the full-network perception linkage relationship topology.

[0132] When constructing the network-wide perception linkage relationship topology, it is necessary to perform similar relationship analysis on each node in the first network-wide perception linkage node set and the second network-wide perception linkage node set. For example, the relationship between the ground base station power consumption node in the first network-wide perception linkage node set and the satellite orbit height node in the second network-wide perception linkage node set. Changes in ground base station power consumption may affect the demand for satellite communications, and thus affect the satellite's orbit adjustment strategy (although this impact may be long-term and complex, involving the synergy of multiple intermediate links and other factors), so the relationship edges between them must also be constructed in the network-wide perception linkage relationship topology.

[0133] Finally, based on the whole network perception linkage relationship topology, the whole network perception data relationship network corresponding to the satellite-ground collaborative network matrix is ​​generated. This whole network perception data relationship network is a comprehensive representation of the complex relationship between the ground-side perception data and the satellite-side status data in the satellite-ground collaborative network matrix.

[0134] It integrates various nodes (including various data nodes of ground base stations and various status data nodes of satellites) and various relationships (direct and indirect) between them in the form of a network. In this network-wide perception data relationship network, each node has its specific meaning and function, and the edges between nodes represent the strength and type of relationship between them. For example, the weight of the edge connecting the ground base station temperature node and the satellite sensor sampling frequency node may represent the degree of influence of this indirect relationship, which can be obtained by analyzing and calculating a large amount of actual data.

[0135] This technical solution of building a network of network-wide perception data relationships based on situational decision information helps to deeply understand the internal structure and operation mechanism of the satellite-ground collaborative network matrix. By clarifying the relationship between the ground network perception data topology and the satellite network status data topology, and building a network-wide perception linkage relationship topology, we can better grasp the mutual influence and collaborative relationship between the ground and satellite data in the satellite-ground collaborative network, thus providing a strong basis for the optimization, management and fault diagnosis of the satellite-ground collaborative network.

[0136] For example, in terms of network optimization, the layout of ground base stations or the orbital parameters of satellites can be adjusted according to the relationship between nodes in the full-network perception data relationship network. If it is found that the relationship between the power consumption nodes of the ground base station and the satellite energy storage nodes has a great impact on the energy consumption of the entire network, the energy efficiency of the entire satellite-ground collaborative network can be improved by optimizing the power consumption management strategy of the ground base station or adjusting the energy distribution method of the satellite. In terms of fault diagnosis, when there is a problem with the network, the nodes and relationships that may have problems can be quickly located through the full-network perception data relationship network. If the number of user connections at the ground base station suddenly drops, by analyzing the full-network perception data relationship network, it can be traced back to the possible problems with the satellite communication link, or the ground base station temperature is too high, affecting the performance of the equipment, so as to carry out targeted troubleshooting and repair.

[0137] With this design, firstly, by using the situation decision information to extract the ground network perception data topology and the satellite network status data topology, the ground-side perception data and satellite-side status data can be accurately and comprehensively characterized, laying a solid foundation for subsequent operations. Secondly, the process of constructing the first and second full-network perception linkage node sets deeply integrates the satellite-ground data topology related information. It is not just a simple node merger, but fully considers the direct and indirect relationship between the satellite and the ground, greatly deepening the connection between the two, so that the understanding of the satellite-ground collaborative network is no longer limited to each isolated part. Furthermore, constructing the full-network perception linkage relationship topology and finally generating the full-network perception data relationship network comprehensively and meticulously presents the complex relationship between satellite and ground data. This comprehensiveness helps to deeply understand the internal structure and operation mechanism of the satellite-ground collaborative network. Finally, this technical solution provides a strong basis for the management, optimization and fault diagnosis of the satellite-ground collaborative network, which is of great significance in improving network performance, optimizing resource allocation, and quickly locating and solving network faults.

[0138] Further, Figure 2 Schematic diagram of a satellite-ground collaborative data construction system 200 provided in an embodiment of the present application. Figure 2 The satellite-ground collaborative data construction system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0139] Alternatively, if Figure 2 As shown, the satellite-ground collaborative data construction system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.

[0140] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0141] Alternatively, if Figure 2 As shown, the satellite-ground collaborative data construction system 200 may also include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices. Specifically, it may send information or data to other devices, or receive information or data sent by other devices.

[0142] Optionally, the satellite-ground collaborative data construction system 200 can implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or the device deployed with the storage engine in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0143] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities.

[0144] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0145] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0146] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0147] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), including a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0148] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the embodiments of the present application are not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the embodiments of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the embodiments of the present application and the scope of protection of the embodiments of the present application, all of which are within the protection of the embodiments of the present application.

Claims

1. A data construction method based on satellite-ground collaborative computing, characterized in that: The method is applied to a satellite-ground collaborative data construction system, and the method includes: Acquire ground-side sensing data corresponding to the satellite-ground collaborative network matrix and multiple groups of satellite-side status data, wherein the ground-side sensing data is ground network equipment monitoring data associated with the satellite-ground collaborative network matrix; Mining ground network embedded coding vectors corresponding to the ground-side sensing data, and mining satellite network embedded coding vectors corresponding to the multiple groups of satellite-side status data respectively; Performing first vector integration on a plurality of satellite network embedded coding vectors and the ground network embedded coding vector respectively, to obtain first satellite-ground network embedded integrated vectors corresponding to the plurality of satellite network embedded coding vectors respectively; Performing a second vector integration on first satellite-to-ground network embedding integration vectors corresponding to the plurality of satellite network embedding coding vectors to obtain a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground collaborative network matrix, wherein the vector feature interaction strategies of the first vector integration and the second vector integration are different; The data construction situation of the satellite-ground collaborative network matrix is ​​judged according to the second satellite-ground network embedded integrated vector to obtain construction situation decision information corresponding to the satellite-ground collaborative network matrix.

2. The method according to claim 1, characterized in that The step of performing first vector integration on the plurality of satellite network embedded coding vectors and the ground network embedded coding vectors to obtain first satellite-ground network embedded integrated vectors corresponding to the plurality of satellite network embedded coding vectors respectively includes: The network state attributes of the multiple satellite network embedded coding vectors are respectively fused with the ground network embedded coding vector to obtain first satellite-ground network embedded integrated vectors corresponding to the multiple satellite network embedded coding vectors.

3. The method according to claim 2, characterized in that Before fusing the plurality of satellite network embedded coding vectors with the ground network embedded coding vectors for network status attributes to obtain first satellite-ground network embedded integrated vectors corresponding to the plurality of satellite network embedded coding vectors, the method further includes: Performing vector pooling processing on the multiple satellite network embedded coding vectors respectively to obtain satellite network embedded pooling vectors corresponding to the multiple satellite network embedded coding vectors respectively, wherein the vector pooling processing is used to compress feature sizes corresponding to the satellite network embedded coding vectors; The step of fusing the plurality of satellite network embedded coding vectors with the ground network embedded coding vectors for network status attributes to obtain first satellite-ground network embedded integrated vectors corresponding to the plurality of satellite network embedded coding vectors respectively includes: The satellite network embedding pooling vectors corresponding to the multiple satellite network embedding coding vectors and the ground network embedding coding vector are fused with network state attributes to obtain first satellite-ground network embedding integrated vectors corresponding to the multiple satellite network embedding coding vectors.

4. The method according to any one of claims 1 to 3, characterized in that: The step of performing a second vector integration on the first satellite-to-ground network embedding integration vectors respectively corresponding to the plurality of satellite network embedding coding vectors to obtain a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordinated network matrix comprises: The first satellite-to-ground network embedding integrated vectors corresponding to the multiple satellite network embedding coding vectors are input into a deep residual model to generate a second satellite-to-ground network embedding integrated vector corresponding to the satellite-to-ground collaborative network matrix.

5. The method according to any one of claims 1 to 3, characterized in that: Before performing second vector integration on the first satellite-to-ground network embedding integration vectors respectively corresponding to the plurality of satellite network embedding coding vectors to obtain the second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordinated network matrix, the method further includes: Acquire at least one first set of satellite-side status data corresponding to a target multi-task information and observation scheduling server from among the multiple sets of satellite-side status data, wherein the target multi-task information and observation scheduling server refers to a multi-task information and observation scheduling server that has a spatiotemporal involvement relationship with the satellite-ground collaborative network matrix; Determine a first satellite-ground coordination confidence coefficient corresponding to the at least one set of first satellite-side status data; Determine a second satellite-ground coordination confidence coefficient corresponding to second satellite-side status data, where the second satellite-side status data refers to satellite-side status data in the multiple groups of satellite-side status data except for the at least one group of first satellite-side status data, and the first satellite-ground coordination confidence coefficient is different from the second satellite-ground coordination confidence coefficient; The step of performing a second vector integration on the first satellite-to-ground network embedding integration vectors respectively corresponding to the plurality of satellite network embedding coding vectors to obtain a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground coordinated network matrix comprises: Performing the first vector integration on the satellite network embedding coding vector and the ground network embedding coding vector corresponding to the at least one set of first satellite side status data according to the first satellite-ground coordination confidence coefficient, to obtain a first satellite-ground network embedding integration vector corresponding to the at least one set of first satellite side status data; The satellite network embedding coding vector and the ground network embedding coding vector corresponding to the second satellite side status data are integrated with the first vector according to the second satellite-ground coordination confidence coefficient to obtain a first satellite-ground network embedding integration vector corresponding to the second satellite side status data.

6. The method according to claim 5, characterized in that The determining of a first satellite-ground coordination confidence coefficient corresponding to the at least one set of first satellite-side status data comprises: Determining the number of groups of the at least one group of first satellite side status data; Determining the probability heat map frequent item features and the probability heat map marginal item features according to the number of groups; According to the timing priority interval of the at least one group of first satellite side status data, the first satellite-ground coordination confidence coefficient is configured for the at least one group of first satellite side status data using a probabilistic thermodynamic statistical rule, wherein the first satellite side status data located within the timing priority interval is configured with the frequent item feature of the probabilistic heat map, and the first satellite side status data located at the boundary of the timing priority interval is configured with the edge item feature of the probabilistic heat map.

7. The method according to any one of claims 1 to 3, characterized in that: The satellite-ground collaborative network matrix includes a plurality of network matrix blocks, wherein the xth network matrix block corresponds to the xth group of satellite side status data, and x is a positive integer; Before mining the satellite network embedding coding vectors corresponding to the multiple groups of satellite side status data, it also includes: determining the yth group of initial satellite side status data in the xth network matrix block as the xth group of satellite side status data, where y is a positive integer.

8. The method according to any one of claims 1 to 3, characterized in that: The step of performing data construction situation determination on the satellite-ground collaborative network matrix according to the second satellite-ground network embedded integrated vector to obtain construction situation decision information corresponding to the satellite-ground collaborative network matrix includes: The second satellite-ground network is embedded in the integrated vector input pre-debugging data to build a situation discrimination network, and generate building situation decision information corresponding to the satellite-ground collaborative network matrix.

9. The method according to claim 8, characterized in that Before embedding the second satellite-to-ground network into the integrated vector input data to build a situation discrimination network and generating the building situation decision information corresponding to the satellite-to-ground collaborative network matrix, the method further includes: Obtain a ground network equipment monitoring data example and multiple sets of satellite side status data examples corresponding to the satellite-ground collaborative network matrix example, wherein the ground network equipment monitoring data example is the ground network equipment monitoring data involved in the data building task process of the satellite-ground collaborative network matrix example, and the satellite-ground collaborative network matrix example carries a building situation decision annotation; Mining the ground network embedded coding vector corresponding to the ground network equipment monitoring data example, and mining the satellite network embedded coding vectors corresponding to the multiple groups of satellite side status data examples respectively; Perform the first vector integration on the plurality of satellite network embedded coding vectors and the ground network embedded coding vector respectively, to obtain first satellite-ground network embedded integrated vectors corresponding to the plurality of groups of satellite side status data examples respectively; Performing the second vector integration on the first satellite-to-ground network embedding integration vectors corresponding to the multiple groups of satellite-side status data examples respectively, to obtain a second satellite-to-ground network embedding integration vector corresponding to the satellite-to-ground collaborative network matrix example; The second satellite-ground network is embedded in the integrated vector input into the initial situation discrimination network to generate the construction situation prediction information corresponding to the satellite-ground collaborative network matrix example; The initial situation discrimination network is debugged based on the difference between the construction situation prediction information and the construction situation decision annotation to obtain the data construction situation discrimination network.

10. A satellite-ground collaborative data construction system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.

Images