A method for intelligent allocation of satellite Internet of Things transmission capacity for disaster prevention

Abstract

The present invention relates to the field of satellite Internet of Things technology, and in particular to a method for intelligently allocating transmission capacity of satellite Internet of Things for disaster prevention, the method specifically comprising: based on satellite status data sets and network quality assessment data, using a network status perception algorithm to generate a network status information table, according to the network status information table, using a graph theory algorithm and a Dijkstra shortest path algorithm to determine different first optimal transmission links between each satellite Internet of Things device and each satellite, and determining a transmission capacity allocation scheme for multiple first optimal transmission links; according to the priority of each data content of ground disaster data, using a resource scheduling algorithm based on deep reinforcement learning, dynamically adjusting the transmission capacity allocation schemes of different first optimal transmission links between each satellite Internet of Things device and each satellite in the disaster-stricken area. The present invention improves the disaster response speed and the reliability of data transmission, and significantly improves the efficiency of disaster warning and response.

Description

Technical Field

[0001] The present invention relates to the technical field of satellite Internet of Things, and in particular to a method for intelligently allocating transmission capacity of a satellite Internet of Things for disaster prevention. Background Art

[0002] With global climate change and frequent natural disasters, disaster prevention and mitigation work has become increasingly important. Traditional disaster prevention methods often rely on ground monitoring equipment and wired communication networks, but these methods are easily damaged when natural disasters occur, resulting in data transmission interruptions and affecting the timeliness and accuracy of emergency response. In order to overcome these limitations, satellite Internet of Things (IoT) technology has shown great application potential in the field of disaster prevention and mitigation due to its wide coverage and strong anti-destruction capabilities.

[0003] However, the existing satellite Internet of Things still faces many challenges in disaster prevention applications, such as limited satellite resources, uneven distribution of data transmission capacity, unstable network status, etc. Especially in the face of multi-type and multi-regional disaster monitoring, how to intelligently and efficiently allocate limited satellite resources to ensure that key data can be transmitted in a timely and accurate manner is an urgent problem to be solved. Summary of the invention

[0004] The purpose of the present invention is to provide a method for intelligently allocating satellite Internet of Things transmission capacity for disaster prevention, which improves the disaster response speed and the reliability of data transmission, significantly improves the efficiency of disaster warning and response, and solves at least one of the above-mentioned prior art problems.

[0005] In a first aspect, the present invention provides a method for intelligently allocating satellite Internet of Things transmission capacity for disaster prevention, the method specifically comprising:

[0006] Obtain historical disaster data for each monitored area, determine the frequency and severity of different disaster types in each monitored area based on the historical disaster data, and determine the data collection ratio of satellite IoT devices used to monitor different disaster types in each monitored area based on the frequency and severity of different disaster types;

[0007] The motion trajectory data of each satellite is obtained according to the prediction of the satellite orbit parameter calculation model, the channel model of each satellite is established through the atmospheric propagation model and the multipath fading model, the energy state data of each satellite is obtained by real-time monitoring, and a satellite state data set is formed according to the motion trajectory data, the parameters of the channel model and the energy state data;

[0008] Based on the data collection ratio, a weighted random sampling algorithm is used to obtain the disaster prediction data collected by each satellite IoT device, and the disaster prediction data is uploaded to the satellite, and network quality assessment data during the uploading process is obtained;

[0009] Based on the satellite status data set and the network quality assessment data, a network status information table is generated by using a network status perception algorithm, and according to the network status information table, different first optimal transmission links between each satellite IoT device and each satellite are determined by using a graph theory algorithm and a Dijkstra shortest path algorithm, and a transmission capacity allocation scheme is determined for the multiple first optimal transmission links;

[0010] When the disaster prediction data of any monitored area reaches the disaster determination threshold, the current monitored area is determined as a disaster-affected area, and ground disaster data of the disaster-affected area is collected by the drone group, and the ground disaster data is uploaded to each satellite;

[0011] When an abnormality occurs during the upload of the ground disaster data, the second optimal transmission link between each UAV and each satellite is recalculated by the Primm minimum spanning tree algorithm, and the transmission rate parameters of each second optimal transmission link are optimized in real time according to the maximum and minimum theorem of network flow;

[0012] According to the priority of each data content of the ground disaster data, a resource scheduling algorithm based on deep reinforcement learning is adopted to dynamically adjust the transmission capacity allocation scheme of different first optimal transmission links between each satellite IoT device in the disaster-stricken area and each satellite.

[0013] In a second aspect, the present invention provides a computer device comprising: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, it implements the satellite Internet of Things transmission capacity intelligent allocation method for disaster prevention as described in any of the above methods.

[0014] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for intelligent allocation of satellite Internet of Things transmission capacity for disaster prevention as described in any of the above methods is implemented.

[0015] Compared with the prior art, the present invention has the following technical effects:

[0016] 1. The present invention effectively integrates satellite Internet of Things technology and artificial intelligence algorithms, improves the speed of disaster response and the reliability of data transmission, and significantly improves the efficiency of disaster warning and response.

[0017] 2. By acquiring the historical disaster data of each monitored area, the present invention can accurately analyze the frequency and severity of different disaster types, thereby scientifically determining the data collection proportion of satellite IoT devices in each monitored area, which helps to reasonably allocate monitoring resources and improve the pertinence and effectiveness of data collection.

[0018] 3. Based on the satellite status data set and network quality assessment data, the present invention adopts network status perception algorithm and graph theory algorithm to determine the optimal transmission link between each satellite IoT device and the satellite, and dynamically adjusts the transmission capacity allocation plan, which can significantly improve data transmission efficiency and reduce transmission delay and packet loss rate.

[0019] 4. When a disaster occurs, the present invention uses a swarm of drones to collect ground disaster data in the disaster-stricken areas, and recalculates the optimal transmission link through the Prim's minimum spanning tree algorithm, and optimizes the transmission rate parameters in real time, which can effectively make up for the shortcomings of satellite Internet of Things equipment in ground data collection and improve the emergency response speed.

[0020] 5. The present invention adopts a resource scheduling algorithm based on deep reinforcement learning to dynamically adjust the transmission capacity allocation plan between satellite IoT devices and satellites in the disaster-stricken areas, which can ensure that key data is transmitted first and provide timely and accurate information support for disaster prevention and mitigation decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a flow chart of a method for intelligently allocating satellite Internet of Things transmission capacity for disaster prevention provided by one embodiment of the present invention;

[0023] Figure 2 It is a structural schematic diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0025] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0026] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0027] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0028] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0029] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0030] In the embodiment of the present application, the execution subject of the process includes a terminal device, which includes but is not limited to: a server, a computer, a smart phone, a tablet computer, and other devices capable of executing the method disclosed in the present application. Figure 1 A schematic diagram of a process flow of a satellite Internet of Things transmission capacity intelligent allocation method for disaster prevention disclosed in an embodiment of the present invention is shown, and the details are as follows:

[0031] S101, obtaining historical disaster data of each monitored area, determining the frequency and severity of different disaster types in each monitored area based on the historical disaster data, and determining the data collection ratio of satellite Internet of Things devices used to monitor different disaster types in each monitored area through the frequency and severity of different disaster types.

[0032] In this embodiment, historical disaster data of each monitoring area is obtained, and the historical disaster data includes information such as the time of occurrence, location of occurrence, affected area, and degree of disaster of different types of disasters in each monitoring area. According to the historical disaster data of each monitoring area obtained, the naive Bayes classification algorithm is used to calculate the frequency and severity of different types of disasters in each monitoring area. The naive Bayes classification algorithm obtains the frequency and severity of different types of disasters in each monitoring area by calculating the prior probability and conditional probability of different types of disasters. On the basis of obtaining the frequency and severity of different types of disasters in each monitoring area, the weighted average method is used to determine the data collection proportion of satellite Internet of Things devices used to monitor different types of disasters in each monitoring area. The weighted average method sets the weight coefficients of different disaster types, and performs weighted average on the frequency and severity of different types of disasters in each monitoring area, so as to obtain the data collection proportion of satellite Internet of Things devices used to monitor different types of disasters in each monitoring area. When determining the data collection ratio of satellite IoT devices used to monitor different types of disasters in each monitoring area, the fuzzy comprehensive evaluation method is used to evaluate the disaster risk level of each monitoring area according to the frequency and severity of different types of disasters in each monitoring area. If the risk level of a certain type of disaster in a certain monitoring area is high, the data collection ratio of satellite IoT devices used to monitor this type of disaster in that monitoring area is increased accordingly. Through the above steps, the data collection ratio of satellite IoT devices used to monitor different types of disasters in each monitoring area is finally determined. According to the determined data collection ratio, the data collection tasks of satellite IoT devices in each monitoring area are allocated to complete the collection of disaster monitoring data in each monitoring area.

[0033] In some embodiments, in the above step S101, determining the occurrence frequency and severity of different disaster types in each monitoring area according to the historical disaster data, and determining the data collection proportion of satellite IoT devices used to monitor different disaster types in each monitoring area according to the occurrence frequency and severity of different disaster types, specifically includes:

[0034] Extracting disaster feature vectors from the historical disaster data, wherein the disaster feature vectors include the occurrence time, location, affected area, and degree of disaster of different disaster types in each monitoring area;

[0035] Based on the disaster feature vector, a naive Bayes classification algorithm is used to calculate the prior probability and conditional probability of different disaster types in each monitoring area, and the occurrence frequency and severity of different disaster types in each monitoring area are obtained;

[0036] Based on the frequency and severity of different disaster types in each monitoring area, the weighted average method is used to determine the first data collection ratio of satellite IoT devices used to monitor different disaster types in each monitoring area;

[0037] Based on the occurrence frequency and severity of different disaster types in each monitoring area, the fuzzy comprehensive evaluation method is used to determine the disaster risk level of different disaster types in each monitoring area, and the second data collection ratio of satellite Internet of Things devices used to monitor different disaster types in each monitoring area is determined according to the disaster risk level;

[0038] The target data collection ratio of satellite Internet of Things devices used to monitor different types of disasters in each monitoring area is determined based on the first data collection ratio and the second data collection ratio.

[0039] In this embodiment, according to the historical disaster data, a data extraction algorithm is used to obtain information such as the time, location, affected area and degree of disaster of different types of disasters in each monitored area, and a disaster feature vector is obtained through feature vector quantization. The vector comprehensively describes the spatiotemporal distribution characteristics and the severity of disasters in each region. The naive Bayes classification algorithm is used to calculate the prior probability and conditional probability of different types of disasters in each monitored area according to the disaster feature vector. The Bayes formula is used to obtain the probability of each type of disaster under given disaster characteristic conditions, and then the occurrence frequency of different disasters in each region is determined. At the same time, combined with the characteristics of the affected area and the degree of disaster, the severity distribution of each type of disaster is obtained.

[0040] According to the historical disaster data of each monitoring area, the occurrence frequency of different disaster types in each area is obtained, and the weight coefficient of each disaster type in each area is determined by the occurrence frequency. According to the historical loss data caused by different disaster types in each monitoring area, the loss amount is used to measure the severity of the disaster, and the average loss amount of each disaster type in each area is obtained. The severity coefficient of each disaster type in each area is determined by the average loss amount. According to the weight coefficient and severity coefficient obtained in the above steps, the weighted average method is used to multiply and normalize the two to obtain the comprehensive weight coefficient of different disaster types in each monitoring area. According to the total number of satellite Internet of Things devices in each monitoring area and the comprehensive weight coefficient of each disaster type, the allocation algorithm is used to allocate satellite Internet of Things devices to each disaster type according to the proportion of the comprehensive weight coefficient, and the number of satellite Internet of Things devices of each disaster type is obtained. According to the number of satellite Internet of Things devices allocated to each disaster type, the division calculation is used to divide the number of devices of each disaster type by the total number of devices to obtain the first data collection proportion of each disaster type.

[0041] The historical occurrence frequency and severity data of different disaster types in each monitoring area are obtained as the input of the fuzzy comprehensive evaluation method. According to the historical data obtained, the membership function in fuzzy mathematics is used to map the occurrence frequency and severity to the interval [0, 1] to obtain the fuzzy evaluation matrix of each disaster type. The weight vectors of the two factors of occurrence frequency and severity are determined by the hierarchical analysis method. Combined with the fuzzy evaluation matrix, the fuzzy synthesis operation is used to obtain the comprehensive evaluation results of different disaster types in each monitoring area. According to the comprehensive evaluation results, the maximum membership principle is used to judge the risk level of different disaster types in each monitoring area. The risk level can be divided into three levels: low risk, medium risk and high risk. According to the determined risk level of different disaster types in each monitoring area, the weighted average method is used, combined with the weights of different risk levels, to calculate the second data collection ratio of satellite Internet of Things devices used to monitor different disaster types in each monitoring area. The obtained second data collection ratio can be used as the data collection configuration parameter of the satellite Internet of Things device in each monitoring area. The data collection strategy of the device is dynamically adjusted according to the ratio to achieve key monitoring of high-risk disaster types.

[0042] For example, a certain region has experienced 5 flood disasters in the past 10 years, with an average affected area of ​​1,000 square kilometers each time and a moderate degree of disaster. The naive Bayes classification algorithm is used to calculate the prior probability of flood disasters in the region as 5, the conditional probability of occurrence in July-September as 8, and the conditional probability of affected area between 800-1,200 square kilometers as 6. It is thus obtained that the frequency of flood disasters in the region is 5 and the severity is moderate. According to the frequency and severity of different disasters in each region, the weighted average method is used to determine the first data collection ratio of satellite IoT devices, and the weight can be set to the sixth power of the frequency and the fourth power of the severity. Then, the fuzzy comprehensive evaluation method is used to construct a disaster risk evaluation matrix, considering the two factors of frequency and severity, to obtain the risk level of different disaster types in each region, and the second data collection ratio is determined accordingly. Finally, the weighted average of the first and second data collection ratios is used to obtain the target data collection ratio, and the weights can be set to 4 and 6 respectively. For example, if the first data collection ratio is 30% and the second data collection ratio is 40%, then the target data collection ratio = 4×30% + 6×40% = 36%.

[0043] S102, obtain the motion trajectory data of each satellite according to the prediction of the satellite orbit parameter calculation model, establish the channel model of each satellite through the atmospheric propagation model and the multipath fading model, obtain the energy status data of each satellite through real-time monitoring, and form a satellite status data set according to the motion trajectory data, the parameters of the channel model and the energy status data.

[0044] In this embodiment, the initial orbital parameters of the satellites over each monitoring area are collected, including the semi-major axis, eccentricity, orbital inclination, right ascension of the ascending node, argument of perigee and mean anomaly, etc. Matlab or similar tools are used in combination with the Kepler problem solving method or the SGP4 model to input the initial orbital parameters of the satellites. The SGP4 model is used to calculate and predict the satellite's motion trajectory data in the future by considering perturbations such as the earth's gravity, atmospheric drag, and the gravity of the sun and the moon.

[0045] According to the satellite's motion trajectory and the characteristics of the atmosphere, a suitable atmospheric propagation model (such as the ITU-R model) is selected to calculate the attenuation of the signal during propagation. Combined with the geometric relationship between the satellite and the ground station, a multipath fading model is constructed to consider the various reflection, refraction and scattering effects that the signal may encounter during propagation. The parameters of the atmospheric propagation model and the multipath fading model are integrated to form a complete channel model, and the channel characteristic parameters (such as attenuation coefficient, phase shift, etc.) are calculated.

[0046] The energy monitoring equipment on the satellite collects the satellite's battery power, solar panel output power and other energy status data in real time. The collected energy status data is processed and analyzed to evaluate the satellite's remaining energy and energy replenishment capacity. Based on the analysis results, an energy early warning mechanism is established to issue an alarm when the satellite's energy is lower than the preset threshold so that timely measures can be taken to ensure the normal operation of the satellite.

[0047] In some embodiments, in the above step S102, the step of obtaining the motion trajectory data of each satellite according to the satellite orbit parameter calculation model prediction specifically includes:

[0048] The motion trajectory data of each satellite is obtained by predicting the satellite orbit parameter calculation model;

[0049] The satellite orbit perturbation model is used to calculate the satellite orbit perturbation force and obtain the satellite orbit perturbation amount;

[0050] Obtaining orbital elements of each satellite, the orbital elements including semi-major axis, eccentricity, orbital inclination, right ascension of ascending node, argument of perigee and mean anomaly;

[0051] The satellite orbit perturbation is added to the satellite orbit element number to obtain the target orbit parameter of each satellite, thereby forming the motion trajectory data of each satellite.

[0052] In this embodiment, according to the satellite orbit parameter calculation model, the orbit element parameters such as orbital elements, eccentricity, orbital inclination, right ascension of ascending node, and argument of perigee of the satellite are obtained. The Kepler equation is used to obtain the position of the satellite in the orbital plane, that is, the true anomaly, through the orbit element parameters. According to the true anomaly, the position of the satellite in the geocentric inertial coordinate system is obtained by using the conversion relationship between the orbital coordinate system and the geocentric inertial coordinate system. Through the J2 perturbation model, the influence of factors such as the earth's flattening on the satellite orbit is obtained, and the satellite position is corrected to obtain a more accurate satellite motion position. According to the relationship between the satellite motion position and time, numerical integration methods such as Runge-Kutta are used to obtain the motion trajectory data of the satellite over a period of time.

[0053] According to the satellite orbit parameters and satellite attitude information, the satellite orbit perturbation model is adopted, and the perturbation factors such as the non-spherical gravity of the earth, atmospheric drag, and solar light pressure are considered to establish the satellite orbit motion equation, and obtain the differential equation group describing the satellite orbit perturbation. The satellite orbit perturbation differential equation group is solved by numerical integration methods, such as the Runge-Kutta method or the Adams-Cowell method, to obtain the numerical solution of the satellite orbit perturbation over a period of time, and determine the change of the satellite orbit under the action of the perturbation force. According to the numerical solution of the satellite orbit perturbation, the orbit perturbation analysis method, such as Fourier analysis or wavelet analysis, is used to extract the main frequency components and amplitude of the satellite orbit perturbation, and determine the main source and influence degree of the satellite orbit perturbation. By performing statistical analysis on the satellite orbit perturbation, the statistical characteristics of the satellite orbit perturbation, such as the mean, variance, and maximum value, are obtained to evaluate the overall impact of the satellite orbit perturbation on the satellite operation.

[0054] The obtained satellite orbit perturbations are added to the corresponding satellite initial orbit elements to obtain the new orbit parameters of each satellite, namely the target orbit parameters. According to the satellite number and the corresponding target orbit parameters, the orbit extrapolation method is used to calculate the detailed orbit information of each satellite within a certain time range, including the position and speed of the satellite, and obtain the motion trajectory data of each satellite over time.

[0055] Specifically, the motion trajectory data of each satellite is obtained according to the prediction of the satellite orbit parameter calculation model. For example, the SGP4 model is used to calculate the satellite orbit, and the satellite orbit height is set to 500km and the inclination is 53°, and the coordinates of the satellite's motion trajectory in one day are calculated. The satellite orbit perturbation model is used to calculate the satellite orbit perturbation force. For example, considering factors such as the non-spherical gravitational perturbation of the earth and the atmospheric resistance perturbation, the satellite orbit perturbation is calculated to be 10m / day. The orbital elements of each satellite are obtained, including the semi-major axis of 6878km, the eccentricity of 001, the orbital inclination of 53°, the ascending node right ascension of 30°, the perigee argument of 60° and the mean anomaly of 0°. The satellite orbit perturbation is added to the satellite orbit elements to obtain the target orbit parameters of each satellite, forming the motion trajectory data of each satellite in the next week.

[0056] In some embodiments, in the above step S102, establishing the channel model of each satellite through the atmospheric propagation model and the multipath fading model specifically includes:

[0057] Obtaining the satellite orbit height and antenna transmission power of each satellite, and obtaining the propagation loss of the satellite signal of each satellite in free space by using a free space propagation loss calculation formula according to the satellite orbit height and the antenna transmission power parameters;

[0058] Obtaining the atmospheric absorption attenuation coefficient of the satellite signal at different frequencies and altitude angles, and obtaining the atmospheric propagation attenuation parameter of the satellite signal;

[0059] According to the number and phase of the paths through which the satellite signal propagates between each satellite IoT device and each satellite, a multipath fading statistical model is used to obtain multipath fading parameters;

[0060] The atmospheric propagation attenuation parameters and multipath fading parameters are substituted into the Rayleigh fading channel model to form a channel model for each satellite.

[0061] In this embodiment, the orbital height of each satellite is obtained according to the orbital parameters of each satellite, and satellite orbital height data is obtained. According to the antenna parameters of each satellite, the antenna transmission power of each satellite is obtained, and satellite antenna transmission power data is obtained. The free space propagation loss calculation formula is adopted, and the satellite orbital height obtained in step 1 and the satellite antenna transmission power obtained in step 2 are used as input parameters to calculate the propagation loss of each satellite signal in free space, and obtain the free space propagation loss data of each satellite signal.

[0062] According to the different frequencies and altitude angles of the satellite signal, the atmospheric absorption attenuation coefficient is obtained. Through the atmospheric absorption attenuation coefficient, the atmospheric propagation attenuation parameter of the satellite signal is obtained. The obtained atmospheric propagation attenuation parameter is used to determine the attenuation of the satellite signal during the propagation process.

[0063] According to the signals received by the satellite IoT device from each satellite, the propagation path number and phase information of each signal are obtained as the input parameters of the multipath fading statistical model. The multipath fading statistical model is used to perform statistical analysis on the propagation path number and phase information of each satellite signal to obtain the statistical characteristic parameters of multipath fading, such as fading depth and duration. By combining the multipath fading statistical characteristic parameters with the antenna gain, receiver sensitivity and other parameters of the satellite IoT device, the multipath fading parameters that comprehensively consider the device performance are obtained. According to the deployment environment of the satellite IoT device, such as cities, suburbs, and rural areas, different multipath fading empirical models are used to correct the multipath fading parameters to obtain multipath fading parameters that are more in line with the actual scenario.

[0064] For example, the atmospheric absorption attenuation coefficient of the satellite signal at a frequency of 2GHz and an altitude angle of 30° is 5dB, and the atmospheric propagation attenuation of the satellite signal is calculated to be -5dB. Based on the two paths of the satellite signal propagating between the satellite IoT device and the satellite and the 30° phase difference, the dual-path Rayleigh fading model is used to calculate the multipath attenuation to be -5dB. Substituting the atmospheric propagation attenuation and multipath attenuation into the Rayleigh fading channel model to form the channel model of each satellite, the total signal attenuation of the satellite communication link is predicted to be -135dB, providing a reference for the design of satellite communication systems.

[0065] S103, based on the data collection ratio, a weighted random sampling algorithm is used to obtain the disaster prediction data collected by each satellite IoT device, and the disaster prediction data is uploaded to the satellite, and network quality assessment data is obtained during the uploading process.

[0066] In this embodiment, according to the data collection ratio of each satellite IoT device, a weighted random sampling algorithm is used to obtain the disaster prediction data collected by each satellite IoT device. The disaster prediction data obtained according to the weighted random sampling algorithm is uploaded to the satellite. In the process of uploading the disaster prediction data, the network quality evaluation data in the uploading process is obtained at the same time, and the network quality evaluation data includes indicators such as network delay, packet loss rate, and bandwidth utilization. By statistically analyzing the network quality evaluation data, the network quality evaluation index is reduced in dimension by the principal component analysis method to obtain a comprehensive evaluation result of the network quality. According to the comprehensive evaluation result of the network quality, it is judged whether the current network quality meets the data upload requirements. If the network quality does not meet the requirements, the adaptive retransmission mechanism is used to re-upload the data that failed to be uploaded until the data is successfully uploaded. The adaptive retransmission mechanism is a data transmission mechanism that dynamically adjusts the number of retransmissions and the retransmission time interval according to the network status, which can effectively improve the reliability of data transmission. Finally, the successfully uploaded disaster prediction data and network quality evaluation data are stored in the satellite database for subsequent disaster prediction and network quality analysis.

[0067] In some embodiments, the above step S103 specifically includes:

[0068] Based on the data collection ratio, determine the weight value of each satellite IoT device;

[0069] A weighted random sampling algorithm is used to determine the data sampling probability of each satellite IoT device according to the weight value, and the disaster prediction data collected by each satellite IoT device is uploaded to each satellite according to the data sampling probability;

[0070] forming first network quality assessment data according to the network transmission condition during the uploading process of the disaster prediction data;

[0071] Based on the principal component analysis method, by calculating the covariance matrix and eigenvalue decomposition, determining a plurality of network quality assessment indicators whose contribution values ​​reach a preset contribution threshold from the first network quality assessment data;

[0072] Determine the weight coefficient of each network quality assessment indicator, and use the weighted average method to calculate the comprehensive network quality assessment score based on multiple network quality assessment indicators and the corresponding weight coefficients to form the target network quality assessment data.

[0073] In this embodiment, according to the data collection ratio of each satellite IoT device, a weighted random sampling algorithm is used to obtain the weight value of each device. The weight value is proportional to the data collection ratio. The larger the data collection ratio, the higher the weight value, and a weight value list of each device is obtained. Through normalization processing, the weight value list is converted into a probability list. The device with a larger weight value has a greater probability of being sampled, and the data sampling probability of each satellite IoT device is determined. According to the data sampling probability, a random number generation method is used to obtain a random number between 0 and 1, and it is determined which device's probability interval the random number falls within. The corresponding device is the selected device, and the selected device needs to upload the collected disaster prediction data to the corresponding satellite. Repeat the execution until the data sampling and uploading of all satellite IoT devices is completed, so that the disaster prediction data distributed on different devices are aggregated to the satellite end.

[0074] According to the upload process of disaster prediction data, various parameters in the network transmission process are obtained, including transmission rate, delay, packet loss rate and other indicators. By statistically analyzing the obtained network transmission parameters, various evaluation indicators reflecting the network transmission quality are obtained, such as average transmission rate, maximum delay time, average packet loss rate, etc. The weighted average method is adopted to weight and sum various network transmission quality evaluation indicators according to their weights of the degree of influence on the upload of disaster prediction data, and obtain a comprehensive network transmission quality score. According to the preset network transmission quality evaluation standard, the comprehensive score is compared with the standard threshold to determine whether the current network transmission quality meets the requirements for uploading disaster prediction data. If the network transmission quality score is lower than the standard threshold, the first network quality evaluation data of unqualified network transmission quality is generated, and the upload process is marked as affected by the network; otherwise, the first network quality evaluation data of qualified network transmission quality is generated.

[0075] According to the first network quality assessment data obtained, the covariance matrix between each network quality assessment index is calculated to measure the correlation between the indexes. The calculated covariance matrix is ​​subjected to eigenvalue decomposition to obtain eigenvalues ​​and corresponding eigenvectors. The eigenvalues ​​represent the variance of each principal component, and the eigenvectors represent the direction of the principal component. According to the eigenvalues, the number of principal components is determined. Usually, the first few principal components whose cumulative contribution rate reaches a preset threshold (such as 80% or 90%) are selected as the final evaluation index. The contribution of each network quality assessment index to the principal component is determined by the element value of the eigenvector. The greater the contribution value, the greater the influence of the index on the principal component, and the higher the importance to the network quality assessment. According to the preset contribution threshold, the network quality assessment index whose contribution value reaches the threshold is screened out as the key indicator for the final evaluation of the communication network quality.

[0076] According to the network quality evaluation index system, the original data of multiple network quality evaluation indicators such as delay, packet loss rate, and bandwidth utilization are obtained. Through data preprocessing and normalization, the standardized values ​​of each network quality evaluation indicator are obtained, which provides a data basis for the subsequent determination of weight coefficients. The hierarchical analysis method is used to compare the network quality evaluation indicators in pairs according to their importance, and a judgment matrix is ​​constructed. The weight coefficients of each network quality evaluation indicator are obtained by calculating the eigenvectors of the judgment matrix. The size of the weight coefficient reflects the contribution of each indicator to the comprehensive evaluation of network quality. According to the standardized values ​​of each network quality evaluation indicator and the corresponding weight coefficient, the weighted average method is used to multiply the standardized values ​​of each indicator by its weight coefficient, and then sum them to obtain the comprehensive network quality evaluation score. This score comprehensively considers the impact of each indicator on the network quality and can fully reflect the overall quality level of the network.

[0077] S104, based on the satellite status data set and the network quality assessment data, a network status perception algorithm is used to generate a network status information table, and according to the network status information table, different first optimal transmission links between each satellite IoT device and each satellite are determined through a graph theory algorithm and a Dijkstra shortest path algorithm, and a transmission capacity allocation plan is determined for multiple first optimal transmission links.

[0078] In this embodiment, based on the satellite status data set (including satellite motion trajectory, channel model parameters, energy status, etc.) and network quality assessment data (such as signal strength, packet loss rate, delay, etc.), a network status perception algorithm is used for analysis. The algorithm comprehensively considers factors such as the distance between satellites, channel quality, satellite remaining energy, and network congestion, and generates a network status information table, which records in detail the real-time status information of the communication link between each satellite IoT device and the satellite.

[0079] Satellite IoT devices and satellites are considered as nodes in graph theory, and the communication links between them are considered as edges. The weight of the edge is set according to the link quality in the network status information table (such as the comprehensive score of factors such as signal strength and packet loss rate). A weighted undirected graph is constructed to represent the communication topology of the entire satellite IoT system.

[0080] The Dijkstra shortest path algorithm is applied to the constructed weighted undirected graph. The algorithm starts from the source node (which can be any satellite IoT device or satellite) and calculates the shortest path to all other nodes in the graph. In disaster prevention applications, it is usually necessary to consider the shortest paths from multiple source nodes to multiple target nodes (such as ground data centers, emergency command centers, etc.). Therefore, the Dijkstra algorithm can be run multiple times, each time starting from a different source node. Based on the results of the Dijkstra algorithm, the first optimal transmission link between each satellite IoT device and each satellite is determined. These links have the best communication performance under the current network status.

[0081] According to the communication capability and business requirements of the first optimal transmission link, a transmission capacity allocation plan is formulated, which needs to ensure that key data (such as high-priority disaster prediction data) can be transmitted preferentially within the limited transmission capacity. A dynamic bandwidth allocation strategy can be adopted to dynamically adjust the transmission capacity of each link according to the real-time network status and business requirements.

[0082] In this embodiment, the transmission capacity allocation scheme based on the network status information table can ensure that resources are reasonably allocated and utilized, avoiding resource waste and congestion. Real-time network status perception and dynamic bandwidth allocation strategy enable the system to adaptively respond to changes in network status and improve the robustness and stability of the system.

[0083] In some embodiments, in the above step S104, the generating of the network status information table by using a network status perception algorithm based on the satellite status data set and the network quality assessment data specifically includes:

[0084] Performing data preprocessing and feature extraction on the satellite state data set to obtain a satellite state feature vector;

[0085] Taking the satellite state feature vector and the network quality assessment data as input, adopting a multi-dimensional data fusion algorithm, and obtaining a network state representation vector through feature weight optimization and dimensionality reduction;

[0086] A graph neural network model is used to learn and infer the network state representation vector, and a full network state information table is obtained through message passing and feature aggregation.

[0087] In this embodiment, according to the satellite state data set, data cleaning and preprocessing methods are used to remove noise, missing values ​​and outliers in the data set, obtain a high-quality satellite state data set, and prepare data for subsequent feature extraction. Through feature engineering methods, key features reflecting the satellite state, such as physical quantities such as the satellite's orbital parameters, attitude angular velocity, and temperature, are extracted from the preprocessed satellite state data set to obtain a feature set that comprehensively describes the satellite state. Feature dimension reduction methods such as principal component analysis are used to remove redundant information in the feature set, obtain a compact and informative satellite state feature subset, and reduce the complexity of subsequent satellite state modeling. According to the feature subset, feature scaling methods such as normalization and standardization are used to map each feature to the same scale, remove the influence of feature dimensions, and obtain a standardized satellite state feature vector.

[0088] The satellite state feature vector and network quality assessment data are used as input, and a multi-dimensional data fusion algorithm is used to determine the importance weight of each dimensional feature through the feature weight optimization method. According to the feature weight optimization result, the dimension reduction technology is used to obtain the network state representation vector after dimensionality reduction through feature selection and feature compression, while retaining the key feature information.

[0089] The satellite state feature vector and network quality assessment data are input into the multi-dimensional data fusion algorithm. Through feature weight optimization and dimensionality reduction technology, redundant and noisy data are removed, key features are extracted, and a compact and representative network state representation vector is obtained. According to the generated network state representation vector, a graph neural network model is constructed, each satellite node is regarded as a node in the graph, and the communication link between satellites is regarded as an edge in the graph, forming the satellite communication network topology. The network state representation vector is learned and inferred through the graph neural network model, and information is transmitted between nodes using the message passing mechanism. The node's own features and neighbor node features are fused through feature aggregation technology to update the node's state representation. After multiple rounds of iterative learning and reasoning, the graph neural network model can capture the global topology of the satellite communication network and the interaction between nodes, and obtain a full network state information table, including the state of each satellite node and the quality of the link between nodes.

[0090] For example, firstly, the satellite status data set is preprocessed and feature extracted, and the minimum-maximum normalization method is used to normalize the data. Ten key features such as satellite orbit height, inclination, and sub-satellite point speed are extracted to form a satellite status feature vector. Then, the satellite status feature vector and five network quality evaluation indicators such as network delay and packet loss rate are used as input, and the multi-view autoencoder algorithm is used to optimize the feature weight and reduce the dimension. By setting the number of hidden layer nodes to 8, the learning rate to 01, and the number of iterations to 100, a 16-dimensional network status representation vector is obtained. Then, the graph neural network model is used to learn and reason about the network status representation vector. By setting the number of graph convolution layers to 3, the number of nodes in each layer to 32, 64, and 128, the maximum pooling is used in the aggregation layer, the activation function is ReLU, and the number of nodes in the output layer is N, an N×N full network status information table is obtained, where N is the total number of satellites and satellite IoT devices.

[0091] In some embodiments, in the above step S104, determining different first optimal transmission links between each satellite IoT device and each satellite by using a graph theory algorithm and a Dijkstra shortest path algorithm according to the network status information table specifically includes:

[0092] According to the network status information table, node identification information of each satellite and each satellite IoT device is obtained, and a vertex set is formed through the node identification information;

[0093] According to the network status information table, a link connection relationship between each satellite and each satellite IoT device is obtained, and an edge set is formed through the link connection relationship;

[0094] Constructing a first satellite Internet of Things network topology graph according to the vertex set and the edge set;

[0095] According to the first satellite Internet of Things network topology diagram, calculate and obtain the distance matrix between each satellite Internet of Things device and each satellite and between each satellite;

[0096] According to the distance matrix, the Dijkstra shortest path algorithm is used to obtain the shortest paths between each satellite IoT device and each satellite and between each satellite, and to determine the first optimal transmission link from each satellite IoT device to each satellite, as well as the first optimal transmission link between satellites.

[0097] In this embodiment, according to the first satellite IoT network topology diagram, a distance calculation formula is used to obtain the longitude and latitude coordinates of each satellite IoT device, and the distance matrix between the satellite IoT devices is obtained by calculating the spherical distance between the devices. According to the first satellite IoT network topology diagram, a distance calculation formula is used to obtain the longitude and latitude coordinates and orbital altitude of each satellite, and the distance matrix between the satellites is obtained by calculating the spatial distance between the satellites.

[0098] According to the distance matrix, the Dijkstra shortest path algorithm is used to obtain the distance information between each satellite IoT device and each satellite and between each satellite to obtain complete distance data. Through the initialization process of the Dijkstra algorithm, the distance from the starting point satellite IoT device to each other satellite is initialized to infinity, and the distance from the starting point to itself is initialized to 0, and the initial state of the algorithm is determined. According to the relaxation operation of the Dijkstra algorithm, all satellite IoT devices and satellite nodes are traversed, and the sum of the distance from the current node to other nodes and the distance from the starting point to other nodes through the current node is compared to update the shortest distance and obtain a better path. The main loop of the Dijkstra algorithm is used to repeat the relaxation operation until all nodes have been visited, and the shortest path from the starting point satellite IoT device to each satellite and between satellites is obtained. According to the obtained shortest path information, the first optimal transmission link from each satellite IoT device to each satellite and the first optimal transmission link from satellite to satellite are determined to obtain the optimal data transmission scheme.

[0099] In some embodiments, in the above step S104, determining a transmission capacity allocation scheme for a plurality of first optimal transmission links specifically includes:

[0100] Establishing a capacity allocation optimization model according to the network status information and the transmission data service demand attributes of each first optimal transmission link, and converting the capacity allocation optimization model into an integer programming model by using an integer programming algorithm;

[0101] The objective function of the integer programming model is to maximize the total capacity that meets the business needs of the satellite Internet of Things, the constraints include link capacity limit and business demand satisfaction, and the decision variables are the transmission capacity allocated to each first optimal transmission link;

[0102] By solving the integer programming model, an optimal transmission capacity allocation scheme for each first optimal transmission link is obtained, and a transmission capacity value allocated to each first optimal transmission link is determined.

[0103] In this embodiment, according to the network status information and transmission data service demand attributes of each first optimal transmission link, a capacity allocation optimization model is established. The model takes the total capacity that maximizes the satellite Internet of Things service demand as the objective function, and considers the link capacity limit and the service demand satisfaction as constraints, and takes the transmission capacity allocated to each first optimal transmission link as a decision variable. The established capacity allocation optimization model is converted into an integer programming model by an integer programming algorithm, so that the decision variable value in the model is limited to an integer, so as to facilitate subsequent solution and practical application. By solving the converted integer programming model, the optimal transmission capacity allocation scheme of each first optimal transmission link is obtained, and the specific transmission capacity value allocated to each first optimal transmission link is determined, so that the total capacity of the satellite Internet of Things service demand is maximized under the constraints of link capacity limit and service demand satisfaction. According to the optimal transmission capacity allocation scheme obtained by solution, the transmission capacity of each first optimal transmission link is reasonably allocated to ensure that different service demands in the satellite Internet of Things system are effectively met, and the overall service quality and user experience of the system are improved. A dynamic adjustment mechanism is adopted to optimize and adjust the transmission capacity allocation plan in a timely manner according to the changes in business needs and network status during the operation of the satellite Internet of Things system, ensuring that the system can continuously and efficiently meet different business needs and maintain stable and reliable transmission services. By real-time monitoring of the actual transmission capacity usage of each first optimal transmission link, it is determined whether there is a capacity bottleneck or idle waste, and the transmission capacity allocation is dynamically optimized based on the monitoring results, thereby improving the resource utilization efficiency and service quality of the satellite Internet of Things system.

[0104] S105, when it is monitored that the disaster prediction data of any monitored area reaches the disaster determination threshold, the current monitored area is determined as a disaster-stricken area, and ground disaster data of the disaster-stricken area is collected according to the drone group, and the ground disaster data is uploaded to each satellite.

[0105] In this embodiment, when it is monitored that the disaster prediction data of any monitored area reaches the disaster determination threshold, the data of the area is analyzed by a multivariate linear regression model to obtain a judgment that the area is a disaster-stricken area. Based on this judgment, a drone group is used to collect ground disaster data in the disaster-stricken area. The data collected by the drone group includes but is not limited to ground damage, affected infrastructure conditions and the distribution of disaster-stricken people. The acquired ground disaster data is uploaded to a low-Earth orbit satellite through wireless transmission technology. After the satellite receives the ground disaster data, it integrates data from different drones through data fusion technology to ensure the integrity and accuracy of the data. The integrated data is transmitted to the ground emergency management center through a satellite communication system. After the emergency management center receives the data, it uses a cluster analysis method to further process the data to identify the most serious disaster areas and the degree of disaster. The processed data is stored in the center's database for subsequent rescue operations and post-disaster reconstruction.

[0106] S106, when an abnormal situation occurs during the uploading process of the ground disaster data, the second optimal transmission link between each UAV and each satellite is recalculated through the Prim's minimum spanning tree algorithm, and the transmission rate parameters of each second optimal transmission link are optimized in real time according to the maximum and minimum theorem of network flow.

[0107] In this embodiment, when the ground disaster data is uploaded to the satellite via the drone, the system monitors the data transmission status in real time, including key indicators such as data packet size, transmission time, packet loss rate, delay, etc. When it is detected that any one or more of these indicators exceeds the preset threshold, the system determines it as an abnormal situation, such as link congestion, signal interference, or drone / satellite failure.

[0108] Once an abnormal situation is detected, the system immediately initiates the emergency response mechanism. First, a new graph theory model is constructed based on the current satellite status dataset (including satellite position, energy status, channel quality, etc.) and drone status (such as position, remaining power, communication module status, etc.). In this model, drones and satellites are used as nodes, and potential communication links between them are used as edges. The weight of the edge is set according to the expected communication quality of the link (considering factors such as signal strength, interference, and expected delay). Applying the Primm minimum spanning tree algorithm, a node is selected from the graph as the starting point, and edges and nodes are gradually added until all nodes are included in the spanning tree with the minimum total weight. This minimum spanning tree represents the optimal set of communication links between drones and satellites under the current network state. In the Primm minimum spanning tree, the paths connecting each drone and satellite are selected as the second optimal transmission links. These links provide the best alternative communication paths under the current abnormal situation.

[0109] According to the network flow maximum-minimum theorem, the system optimizes the transmission rate parameters of the second-best transmission link in real time. The theorem states that in a network, the maximum flow from the source to the sink is equal to the capacity of the path with the minimum capacity among all possible paths. The system first calculates the maximum available bandwidth of each second-best transmission link, and then dynamically adjusts the transmission rate of each link based on the flow demand and link capacity limit of the entire network to ensure that the overall throughput of the network is maximized while meeting all transmission requirements.

[0110] In this embodiment, through real-time monitoring and anomaly detection mechanisms, the system can quickly identify problems in the data transmission process and immediately initiate an emergency response process to reduce the data transmission interruption time caused by abnormal situations. The application of Prim's minimum spanning tree algorithm ensures that in abnormal situations, the optimal alternative communication link can be found between the drone and the satellite to ensure the continuity and reliability of data transmission. The transmission rate parameter optimization based on the maximum and minimum theorem of network flow can ensure the rational allocation and utilization of network resources, avoid resource waste and congestion, and improve the transmission efficiency of the overall network.

[0111] In some embodiments, in the above step S106, the second optimal transmission link between each UAV and each satellite is recalculated by the Prim's minimum spanning tree algorithm, and the transmission rate parameters of each second optimal transmission link are optimized in real time according to the maximum and minimum theorem of network flow, specifically including:

[0112] According to the communication link status information between each drone and each satellite, a second satellite Internet of Things network topology graph is constructed, wherein the nodes of the second satellite Internet of Things network topology graph are each drone and each satellite, the edges of the second satellite Internet of Things network topology graph are the communication links between the drone and the satellite, and the weights of the edges of the second satellite Internet of Things network topology graph are the transmission rates of the communication links between the drone and the satellite;

[0113] Using Prim's minimum spanning tree algorithm to construct a minimum spanning tree for the second satellite Internet of Things network topology graph to obtain a second optimal transmission link;

[0114] When an abnormality occurs in the second optimal transmission link, an edge with the smallest weight is selected from the remaining unselected edges of the second satellite IoT network topology graph, and the nodes connected to the edge are added to the minimum spanning tree until all nodes are added to the minimum spanning tree, thereby obtaining a reconstructed second optimal transmission link;

[0115] By adopting the maximum-minimum theorem of network flow and analyzing the relationship between the transmission rate parameters of each second-best transmission link and the service demand, the transmission link rate parameter optimization objective function is obtained;

[0116] Using a linear programming algorithm, by solving the optimal solution of the transmission link rate parameter optimization objective function, the optimal value of the transmission rate parameter of each second optimal transmission link is obtained;

[0117] According to the optimal value of the transmission rate parameter of each second optimal transmission link, the transmission rate configuration of each second optimal transmission link is dynamically adjusted.

[0118] In this embodiment, by setting the initial parameters of the Prim minimum spanning tree algorithm, including the starting node, the minimum spanning tree set and the candidate edge set, etc., the initial conditions required by the algorithm are obtained to prepare for the subsequent construction of the minimum spanning tree. The Prim minimum spanning tree algorithm is adopted, starting from the starting node, and the edge with the smallest weight is continuously selected from the candidate edge set to be added to the minimum spanning tree set, and the candidate edge set is updated at the same time until all nodes are added to the minimum spanning tree set, and the minimum spanning tree of the second satellite Internet of Things network topology map is obtained. According to the constructed minimum spanning tree, the optimal transmission link between the drone and the satellite is extracted, the second optimal transmission link is obtained, and the optimal path for transmitting data between the drone and the satellite is determined. By analyzing and evaluating the second optimal transmission link, it is determined whether the transmission rate, delay, reliability and other indicators of the transmission link meet the system requirements, and optimization and adjustment are performed according to the actual situation to obtain the final second optimal transmission link solution.

[0119] According to the maximum and minimum theorem of network flow, the transmission rate parameters of each second optimal transmission link are obtained, and the transmission link rate parameter optimization objective function is obtained by analyzing the relationship between the transmission rate parameters and business requirements. According to the transmission link rate parameter optimization objective function, a linear programming algorithm is used to obtain the optimal value of the transmission rate parameter of each second optimal transmission link by solving the optimal solution of the objective function. According to the business attributes of each second optimal transmission link, the simplex method in the linear programming algorithm is used to obtain the objective function and constraint conditions by establishing a linear programming model. According to the objective function and constraint conditions, the iterative calculation process of the simplex method is used to continuously search for feasible solutions and optimal solutions to obtain the optimal value of the transmission rate parameter of each second optimal transmission link.

[0120] Assume that there are 5 drones and 3 satellites, and the transmission rates of the communication links between them are: 10Mbps between drone 1 and satellite 1, 8Mbps between drone 1 and satellite 2, 12Mbps between drone 2 and satellite 1, 9Mbps between drone 2 and satellite 3, 11Mbps between drone 3 and satellite 2, 7Mbps between drone 4 and satellite 1, 13Mbps between drone 4 and satellite 3, and 6Mbps between drone 5 and satellite 2. Based on this information, a network topology graph containing 8 nodes (5 drones and 3 satellites) and 8 edges can be constructed, where the weight of the edge is the transmission rate of the corresponding communication link. The minimum spanning tree of the network topology graph is constructed using the Prim minimum spanning tree algorithm. First, the edge with the smallest weight (the edge between drone 5 and satellite 2) is selected, and the nodes connected to it are added to the minimum spanning tree; then, among the remaining unselected edges, the edge with the smallest weight (the edge between drone 4 and satellite 1) is selected, and the nodes connected to it are added to the minimum spanning tree; and so on, until all nodes are added to the minimum spanning tree, and finally the second optimal transmission link is obtained. When an abnormality occurs in the second optimal transmission link, such as the link between UAV 4 and satellite 1 is interrupted, the edge with the smallest weight (the edge between UAV 1 and satellite 2) is selected from the remaining unselected edges, and the nodes connected to it are added to the minimum spanning tree to reconstruct the second optimal transmission link. The network flow maximum and minimum theorem is used to analyze the relationship between the transmission rate parameters of each second optimal transmission link and the service requirements, and the transmission link rate parameter optimization objective function is obtained, such as maximizing network throughput and minimizing transmission delay. The linear programming algorithm is used to solve the optimal solution of the transmission link rate parameter optimization objective function, and the optimal value of the transmission rate parameter of each second optimal transmission link is obtained, such as adjusting the link transmission rate between UAV 1 and satellite 2 to 9Mbps, and the link transmission rate between UAV 4 and satellite 3 to 11Mbps. Finally, according to the obtained optimal value of the transmission rate parameter, the transmission rate configuration of each second optimal transmission link is dynamically adjusted to ensure the optimal network performance.

[0121] S107, according to the priority of each data content of the ground disaster data, a resource scheduling algorithm based on deep reinforcement learning is adopted to dynamically adjust the transmission capacity allocation scheme of different first optimal transmission links between each satellite IoT device in the disaster-stricken area and each satellite.

[0122] In this embodiment, first, the priority of each data content of the ground disaster data is evaluated. This can be determined based on the urgency of the data, the direct impact on disaster response, and the needs of subsequent analysis and processing. For example, real-time disaster images, damage to key infrastructure, casualty reports, etc. may have a higher priority.

[0123] Then, a deep reinforcement learning environment is constructed, which includes a state space, an action space, a reward function, and a deep learning model (such as a convolutional neural network (CNN) combined with a recurrent neural network (RNN). The state space can include the current network state (such as the remaining capacity of each link, transmission delay, and packet loss rate), the state of the satellite IoT device (such as energy level and processing power), and the priority of the data content. The action space corresponds to the possible transmission capacity allocation scheme, that is, how to dynamically adjust the transmission capacity of each first optimal transmission link. The reward function is designed to give positive or negative feedback based on the timeliness, accuracy, and utilization efficiency of network resources of data transmission. For example, when high-priority data is transmitted quickly and the overall network load is balanced, the reward value is high; on the contrary, if low-priority data occupies too many resources and causes high-priority data to be delayed, the reward value is low. The deep reinforcement learning model is trained using historical data or simulated data. During the training process, the model optimizes its decision-making strategy by continuously trying different transmission capacity allocation schemes and obtaining feedback based on the reward function. As the training progresses, the model gradually learns how to make the best transmission capacity allocation decision under different states.

[0124] In actual operation, when new ground disaster data is received, the system first evaluates its priority and inputs the current network status and satellite IoT device status into the trained deep reinforcement learning model. The model calculates the optimal transmission capacity allocation plan based on the input information and adjusts the transmission capacity of each first optimal transmission link in real time.

[0125] In this embodiment, the resource scheduling algorithm based on deep reinforcement learning can intelligently and dynamically adjust the transmission capacity allocation plan according to the priority of the data content and the network status, ensuring that high-priority data is transmitted first.

[0126] In some embodiments, the above step S107 specifically includes:

[0127] According to the priority of each data content of the ground disaster data, a resource scheduling model is constructed using a deep Q network algorithm, and the resource scheduling model includes a state space, an action space and a reward function;

[0128] The state space includes the data transmission requirements of each satellite IoT device, the orbital position and communication capacity of each satellite, the action space includes the transmission capacity allocation scheme for each first optimal transmission link, and the reward function includes the data transmission priority, delay and packet loss rate of each first optimal transmission link;

[0129] In the process of training the resource scheduling model, action selection is performed according to the state space and the action space using an ε-greedy strategy, the reward function is updated using a temporal difference algorithm, and training is iterated until the network converges to obtain a transmission capacity allocation plan for each first optimal transmission link.

[0130] In this embodiment, according to the priority of each data content of the ground disaster data, a deep Q network algorithm is used to construct a resource scheduling model, which includes a state space, an action space and a reward function, wherein the state space includes the data transmission requirements of each satellite IoT device, the orbital position and communication capacity of each satellite, the action space includes the transmission capacity allocation scheme for each first optimal transmission link, and the reward function includes the data transmission priority, delay and packet loss rate of each first optimal transmission link. The state information such as the data transmission requirements of each satellite IoT device, the orbital position and communication capacity of each satellite are obtained, and these state information are input into the network through the input layer of the deep Q network to obtain the Q value in the current state. The ε-greedy strategy is used to select the action, that is, an action is randomly selected with a probability of ε, otherwise the action with the largest Q value is selected, which corresponds to the transmission capacity allocation scheme for each first optimal transmission link, and the transmission capacity of each link is determined. According to the selected action, the feedback information such as the data transmission priority, delay and packet loss rate of each first optimal transmission link is obtained, and the Q value under the action is calculated through the output layer of the deep Q network, and compared with the target Q value to obtain the time difference error. The parameters of the deep Q network are updated by the stochastic gradient descent method, so that the output Q value of the network is constantly close to the target Q value. Through multiple iterative training, the network converges and the optimal resource scheduling model is obtained. The trained resource scheduling model is applied to the actual satellite IoT data transmission scenario. According to the data transmission requirements of each satellite IoT device and the state information such as the orbital position of the satellite, the trained model is used to make decisions to obtain the optimal transmission link and transmission capacity allocation plan, thereby achieving efficient and reliable data transmission.

[0131] For example, at a certain moment, the data transmission demand of satellite A is 10Mbps, the orbital position is (120°E, 45°N), and the communication capacity is 100Mbps. The action space includes the transmission capacity allocation scheme for each first optimal transmission link, such as 20Mbps for link 1 and 30Mbps for link 2. The reward function comprehensively considers the data transmission priority, delay and packet loss rate of each first optimal transmission link, and is designed as R=5×priority+3×(1-delay)+2×(1-packet loss rate). During the model training process, the action selection is performed according to the ε-greedy strategy, that is, the action with the largest current Q value is selected with a probability of 1-ε, and the action is randomly selected with a probability of ε, where ε is 1. The reward function is updated using the temporal difference algorithm, that is, Q(s, a)←Q(s, a)+α, where α is the learning rate, which is 01, and γ is the discount factor, which is 9. The training is continuously iterated until the reward function converges, and the transmission capacity allocation scheme for each first optimal transmission link is finally obtained. Through this solution, the transmission delay and packet loss rate can be minimized while ensuring the data transmission priority, thereby achieving efficient utilization of satellite communication resources.

[0132] Reference Figure 2 The embodiment of the present invention further provides a computer device 2, comprising: a memory 202 and a processor 201 and a computer program 203 stored in the memory 202. When the computer program 203 is executed on the processor 201, the satellite Internet of Things transmission capacity intelligent allocation method for disaster prevention as described in any one of the above methods is implemented.

[0133] The computer device 2 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 2 may include, but is not limited to, a processor 201 and a memory 202. Those skilled in the art will appreciate that Figure 2 It is only an example of the computer device 2 and does not constitute a limitation on the computer device 2. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, it may also include input and output devices, network access devices, etc.

[0134] The processor 201 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0135] In some embodiments, the memory 202 may be an internal storage unit of the computer device 2, such as a hard disk or memory of the computer device 2. In other embodiments, the memory 202 may also be an external storage device of the computer device 2, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 2. Further, the memory 202 may also include both an internal storage unit of the computer device 2 and an external storage device. The memory 202 is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory 202 may also be used to temporarily store data that has been output or is to be output.

[0136] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for intelligent allocation of satellite Internet of Things transmission capacity for disaster prevention as described in any of the above methods is implemented.

[0137] In this embodiment, if the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0138] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0139] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0140] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0141] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Claims

1. A method for intelligent allocation of satellite Internet of Things transmission capacity for disaster prevention, characterized in that: The method specifically comprises: Obtain historical disaster data for each monitored area, determine the frequency and severity of different disaster types in each monitored area based on the historical disaster data, and determine the data collection ratio of satellite IoT devices used to monitor different disaster types in each monitored area based on the frequency and severity of different disaster types; The motion trajectory data of each satellite is obtained according to the prediction of the satellite orbit parameter calculation model, the channel model of each satellite is established through the atmospheric propagation model and the multipath fading model, the energy state data of each satellite is obtained by real-time monitoring, and a satellite state data set is formed according to the motion trajectory data, the parameters of the channel model and the energy state data; Based on the data collection ratio, a weighted random sampling algorithm is used to obtain the disaster prediction data collected by each satellite IoT device, and the disaster prediction data is uploaded to the satellite, and network quality assessment data during the uploading process is obtained; Based on the satellite status data set and the network quality assessment data, a network status information table is generated by using a network status perception algorithm, and according to the network status information table, different first optimal transmission links between each satellite IoT device and each satellite are determined by using a graph theory algorithm and a Dijkstra shortest path algorithm, and a transmission capacity allocation scheme is determined for the multiple first optimal transmission links; When the disaster prediction data of any monitored area reaches the disaster determination threshold, the current monitored area is determined as a disaster-affected area, and a group of drones are used to collect ground disaster data of the disaster-affected area, and the ground disaster data is uploaded to each satellite; When an abnormality occurs during the upload of the ground disaster data, the second optimal transmission link between each UAV and each satellite is recalculated by the Primm minimum spanning tree algorithm, and the transmission rate parameters of each second optimal transmission link are optimized in real time according to the maximum and minimum theorem of network flow; According to the priority of each data content of the ground disaster data, a resource scheduling algorithm based on deep reinforcement learning is adopted to dynamically adjust the transmission capacity allocation scheme of different first optimal transmission links between each satellite IoT device in the disaster-stricken area and each satellite.

2. The method according to claim 1, characterized in that Determining the occurrence frequency and severity of different disaster types in each monitoring area according to the historical disaster data, and determining the data collection proportion of satellite IoT devices used to monitor different disaster types in each monitoring area according to the occurrence frequency and severity of different disaster types, specifically includes: Extracting disaster feature vectors from the historical disaster data, wherein the disaster feature vectors include the occurrence time, occurrence location, disaster-affected area and disaster severity of different disaster types in each monitoring area; Based on the disaster feature vector, a naive Bayes classification algorithm is used to calculate the prior probability and conditional probability of different disaster types in each monitoring area, and the occurrence frequency and severity of different disaster types in each monitoring area are obtained; Based on the frequency and severity of different disaster types in each monitoring area, the weighted average method is used to determine the first data collection ratio of satellite IoT devices used to monitor different disaster types in each monitoring area; Based on the occurrence frequency and severity of different disaster types in each monitoring area, the fuzzy comprehensive evaluation method is used to determine the disaster risk level of different disaster types in each monitoring area, and the second data collection ratio of satellite Internet of Things devices used to monitor different disaster types in each monitoring area is determined according to the disaster risk level; The target data collection ratio of satellite Internet of Things devices used to monitor different types of disasters in each monitoring area is determined based on the first data collection ratio and the second data collection ratio.

3. The method according to claim 1, characterized in that The step of obtaining the motion trajectory data of each satellite according to the satellite orbit parameter calculation model prediction specifically includes: The satellite orbit perturbation model is used to calculate the satellite orbit perturbation force and obtain the satellite orbit perturbation amount; Obtaining orbital elements of each satellite, the orbital elements including semi-major axis, eccentricity, orbital inclination, right ascension of ascending node, argument of perigee and mean anomaly; The satellite orbit perturbation is added to the satellite orbit element number to obtain the target orbit parameter of each satellite, thereby forming the motion trajectory data of each satellite.

4. The method according to claim 1, characterized in that: The channel model of each satellite is established by using the atmospheric propagation model and the multipath fading model, specifically including: Obtaining satellite orbit height and antenna transmission power parameters of each satellite, and obtaining the propagation loss of satellite signals of each satellite in free space using a free space propagation loss calculation formula according to the satellite orbit height and the antenna transmission power parameters; Obtaining the atmospheric absorption attenuation coefficient of the satellite signal at different frequencies and altitude angles, and obtaining the atmospheric propagation attenuation parameter of the satellite signal; According to the number and phase of the paths through which the satellite signal propagates between each satellite IoT device and each satellite, a multipath fading statistical model is used to obtain multipath fading parameters; The atmospheric propagation attenuation parameters and multipath fading parameters are substituted into the Rayleigh fading channel model to form a channel model for each satellite.

5. The method according to claim 1, characterized in that Based on the data collection ratio, a weighted random sampling algorithm is used to obtain the disaster prediction data collected by each satellite IoT device, and the disaster prediction data is uploaded to the satellite, and network quality evaluation data is obtained during the uploading process, specifically including: Based on the data collection ratio, determine the weight value of each satellite IoT device; A weighted random sampling algorithm is used to determine the data sampling probability of each satellite IoT device according to the weight value, and the disaster prediction data collected by each satellite IoT device is uploaded to each satellite according to the data sampling probability; forming first network quality assessment data according to the network transmission condition during the uploading process of the disaster prediction data; Based on the principal component analysis method, by calculating the covariance matrix and eigenvalue decomposition, determining a plurality of network quality assessment indicators whose contribution values ​​reach a preset contribution threshold from the first network quality assessment data; Determine the weight coefficient of each network quality assessment indicator, and use the weighted average method to calculate the comprehensive network quality assessment score based on multiple network quality assessment indicators and the corresponding weight coefficients to form the target network quality assessment data.

6. The method according to claim 1, characterized in that The generating of the network status information table by using a network status perception algorithm based on the satellite status data set and the network quality assessment data specifically includes: Performing data preprocessing and feature extraction on the satellite state data set to obtain a satellite state feature vector; Taking the satellite state feature vector and the network quality assessment data as input, adopting a multi-dimensional data fusion algorithm, and obtaining a network state representation vector through feature weight optimization and dimensionality reduction; A graph neural network model is used to learn and infer the network state representation vector, and a full network state information table is obtained through message passing and feature aggregation.

7. The method according to claim 1, characterized in that Determining different first optimal transmission links between each satellite IoT device and each satellite by using a graph theory algorithm and a Dijkstra shortest path algorithm according to the network status information table specifically includes: According to the network status information table, node identification information of each satellite and each satellite IoT device is obtained, and a vertex set is formed through the node identification information; According to the network status information table, a link connection relationship between each satellite and each satellite IoT device is obtained, and an edge set is formed through the link connection relationship; Constructing a first satellite Internet of Things network topology graph according to the vertex set and the edge set; According to the first satellite Internet of Things network topology diagram, calculate and obtain the distance matrix between each satellite Internet of Things device and each satellite and between each satellite; According to the distance matrix, the Dijkstra shortest path algorithm is used to obtain the shortest paths between each satellite IoT device and each satellite and between each satellite, and to determine the first optimal transmission link from each satellite IoT device to each satellite, as well as the first optimal transmission link between satellites.

8. The method according to claim 1, characterized in that The step of determining a transmission capacity allocation scheme for the plurality of first optimal transmission links specifically includes: Establishing a capacity allocation optimization model according to the network status information and the transmission data service demand attributes of each first optimal transmission link, and converting the capacity allocation optimization model into an integer programming model by using an integer programming algorithm; The objective function of the integer programming model is to maximize the total capacity that meets the business needs of the satellite Internet of Things, the constraints include link capacity limit and business demand satisfaction, and the decision variables are the transmission capacity allocated to each first optimal transmission link; By solving the integer programming model, an optimal transmission capacity allocation scheme for each first optimal transmission link is obtained, and a transmission capacity value allocated to each first optimal transmission link is determined.

9. The method according to claim 1, characterized in that: The method of recalculating the second optimal transmission link between each UAV and each satellite by using the Prim minimum spanning tree algorithm and optimizing the transmission rate parameters of each second optimal transmission link in real time according to the maximum and minimum theorem of network flow specifically includes: According to the communication link status information between each drone and each satellite, a second satellite Internet of Things network topology graph is constructed, wherein the nodes of the second satellite Internet of Things network topology graph are each drone and each satellite, the edges of the second satellite Internet of Things network topology graph are the communication links between the drone and the satellite, and the weights of the edges of the second satellite Internet of Things network topology graph are the transmission rates of the communication links between the drone and the satellite; Using Prim's minimum spanning tree algorithm to construct a minimum spanning tree for the second satellite Internet of Things network topology graph to obtain a second optimal transmission link; When an abnormality occurs in the second optimal transmission link, an edge with the smallest weight is selected from the remaining unselected edges of the second satellite IoT network topology graph, and the nodes connected to the edge are added to the minimum spanning tree until all nodes are added to the minimum spanning tree, thereby obtaining a reconstructed second optimal transmission link; By adopting the network flow maximum and minimum theorem, the transmission link rate parameter optimization objective function is obtained by analyzing the relationship between the transmission rate parameters of each second optimal transmission link and the service demand; Using a linear programming algorithm, by solving the optimal solution of the transmission link rate parameter optimization objective function, the optimal value of the transmission rate parameter of each second optimal transmission link is obtained; According to the optimal value of the transmission rate parameter of each second optimal transmission link, the transmission rate configuration of each second optimal transmission link is dynamically adjusted.

10. The method according to claim 1, characterized in that The method of dynamically adjusting the transmission capacity allocation scheme of different first optimal transmission links between each satellite IoT device and each satellite in the disaster-stricken area according to the priority of each data content of the ground disaster data and using a resource scheduling algorithm based on deep reinforcement learning specifically includes: According to the priority of each data content of the ground disaster data, a resource scheduling model is constructed using a deep Q network algorithm, and the resource scheduling model includes a state space, an action space and a reward function; The state space includes the data transmission requirements of each satellite IoT device, the orbital position and communication capacity of each satellite, the action space includes the transmission capacity allocation scheme for each first optimal transmission link, and the reward function includes the data transmission priority, delay and packet loss rate of each first optimal transmission link; In the process of training the resource scheduling model, action selection is performed according to the state space and the action space using an ε-greedy strategy, the reward function is updated using a temporal difference algorithm, and training is iterated until the network converges to obtain a transmission capacity allocation plan for each first optimal transmission link.

Images