Energy saving routing method and system in low earth orbit satellite constellation

Abstract

The application relates to the technical field of satellite communication, and particularly discloses an energy-saving routing method and system in a low-orbit satellite constellation, which comprises the following steps: constructing a system model containing a satellite constellation topology, satellite energy consumption, a solar eclipse period determination and a task flow, taking minimization of satellite battery life loss as a core optimization target, and establishing a joint optimization problem; modeling the joint optimization problem into a Markov decision process in a limited time domain based on a battery attenuation model and dynamic channel sensing; solving the Markov decision process by adopting a double-layer optimization framework which fuses a Q-learning algorithm and a reverse inductive battery management algorithm; and outputting a final communication resource configuration and task scheduling strategy based on a solving result. The application can obtain an optimal communication resource configuration and task scheduling strategy, effectively prolongs satellite battery life, reduces a dynamic energy consumption range, guarantees task QoS requirements, and improves the operation stability and economy of a large-scale multi-orbit LEO satellite constellation.

Description

An energy-saving routing method and system for low-Earth orbit satellite constellations Technical Field

[0001] This invention relates to the field of satellite communication technology, specifically to an energy-saving routing method and system for low-Earth orbit satellite constellations. Background Technology

[0002] With the diversification of satellite communication needs, low-Earth orbit (LEO) satellite constellations are widely used in urban IoT management, disaster emergency communication, and global transportation infrastructure monitoring due to their advantages of low latency and wide coverage. LEO satellite constellations typically consist of hundreds to thousands of satellites, relying on solar power and battery storage for operation. However, satellite batteries have limited lifespan due to charge-discharge cycles, especially deep discharge (DOD), which significantly accelerates battery degradation. Current LEO satellite communication routing decisions are primarily based on communication latency and link capacity, without fully considering satellite battery status. When satellites are in shadow, they rely entirely on battery power. Continuously processing communication requests at a fixed power level leads to deep battery discharge, shortening battery life. Satellite battery replacement and satellite launch costs are extremely high, and frequent replacements significantly increase the operating costs of large-scale constellations, even impacting their commercial viability. Existing technologies suffer from the following drawbacks: they fail to balance mission quality of service (QoS) with battery life, resulting in excessive wear and tear on some satellites and shortening the overall constellation's service life; they lack dynamic awareness of battery status and linkage with routing strategies, allowing satellites with severely degraded batteries to still bear a heavy communication load; and routing optimization does not consider the randomness of solar charging, making it difficult to cope with energy supply and demand fluctuations in shadow and sunlight areas. Summary of the Invention

[0003] The purpose of this invention is to provide an energy-saving routing method and system for low-Earth orbit satellite constellations to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] An energy-efficient routing method in a low-Earth orbit satellite constellation, the method comprising:

[0006] A system model is constructed that includes satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic. The core optimization objective is to minimize satellite battery life loss, and a joint optimization problem is established.

[0007] Based on the battery attenuation model and dynamic channel awareness, the joint optimization problem is modeled as a Markov decision process in the finite time domain. Its elements include a state space defined by the task queue, channel quality and track coordination state, an action space defined by transmission power, antenna gain and task operation switch, and a reward function that includes battery life penalty and QoS reward.

[0008] A two-layer optimization framework integrating Q-learning algorithm and back-induction battery management algorithm is adopted to solve the Markov decision process. The upper layer dynamically optimizes the communication resource configuration parameters through Q-learning, and the lower layer optimizes the task scheduling sequence through back-induction.

[0009] Based on the solution results, combined with cross-track link cost calculation and dynamic feedback mechanism, the final communication resource configuration and task scheduling strategy are output, and adaptive adjustments are made according to real-time battery status and channel quality.

[0010] As a further aspect of the present invention, the satellite energy consumption model quantifies the energy consumption of the satellite in each stage of transmission, reception, nominal operation and data processing, and defines the dynamic update equation of battery energy and the quantitative equation of battery life loss related to the depth of discharge.

[0011] A solar eclipse period determination model is established based on the spatial geometric relationship between the Earth, the Sun and the satellite, which is used to identify in real time whether the satellite is in the sunshine period, the umbra or the penumbra.

[0012] A task traffic model based on geographical region division is established to generate a traffic matrix that varies with space and time. The strength of the traffic matrix is ​​positively correlated with the regional user density and the number of terminals, and negatively correlated with the distance between regions.

[0013] As a further embodiment of the present invention, the objective function of the joint optimization problem is:

[0014] ;

[0015] in, and For weighting coefficients, Qos total To normalize the overall QoS index, This represents the average battery life consumption of all satellites during the time period from t1 to t2.

[0016] As a further embodiment of the present invention, in the two-layer optimization framework, the Q-value update of the upper-layer Q-learning algorithm follows the following rules:

[0017] ;

[0018] Where α is the learning rate, γ is the discount factor, and the Q-table is initialized as follows: .

[0019] As a further aspect of the present invention, in the lower layer, the task scheduling sequence is optimized by back-induction, and the value function based on the Bellman equation is derived, and the optimal action sequence is solved by back-induction iteration.

[0020] As a further embodiment of the present invention, the formula for calculating the cost of the inter-track link is as follows:

[0021] ;

[0022] in, Let x be the difference in orbital altitude between the satellite and y be the ... For propagation loss, Here, λ is the orbital weighting coefficient, and λ is the transmission wavelength. L p Losses due to rainfall.

[0023] As a further embodiment of the present invention, the dynamic feedback mechanism includes:

[0024] Battery safety dynamic constraints adjust the maximum permissible depth of discharge in real time based on the battery health status.

[0025] Channel adaptation constraint: When the real-time channel quality is lower than a preset threshold, it triggers compensation adjustment of antenna gain or transmission power.

[0026] Dynamic QoS constraints for tasks: The minimum transmission rate or maximum latency requirement is dynamically set based on task priority.

[0027] The present invention also provides an energy-saving routing system in a low-Earth orbit satellite constellation, the system comprising:

[0028] The optimization module is used to construct a system model that includes satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic, with the core optimization objective of minimizing satellite battery life loss, and to establish a joint optimization problem.

[0029] The modeling module is used to model the joint optimization problem as a finite-time Markov decision process based on the battery attenuation model and dynamic channel awareness. Its elements include a state space defined by the task queue, channel quality and track coordination state, an action space defined by transmission power, antenna gain and task operation switch, and a reward function that includes battery life penalty and QoS reward.

[0030] The solution module is used to solve the Markov decision process using a two-layer optimization framework that integrates the Q-learning algorithm and the back-inductive battery management algorithm. The upper layer dynamically optimizes the communication resource configuration parameters through Q-learning, and the lower layer optimizes the task scheduling sequence through back-induction.

[0031] The strategy output module is used to output the final communication resource configuration and task scheduling strategy based on the solution results, combined with cross-track link cost calculation and dynamic feedback mechanism, and to adaptively adjust it according to real-time battery status and channel quality.

[0032] Compared with the prior art, the beneficial effects of the present invention are: through precise multi-factor coupling modeling and two-layer optimization strategy, the battery degradation caused by ineffective charging and discharging is reduced, the battery life consumption rate is reduced, and the satellite's on-orbit service life is extended.

[0033] By quantifying the relationship between mission QoS and satellite lifetime, we can ensure the differentiated service needs of missions with different priorities and optimize the overall service performance of the system.

[0034] It can better cope with the impact of channel time-varying characteristics, reduce the dynamic power consumption range, improve the ability to resist channel fading and interference, and maintain stable communication quality;

[0035] The multi-track collaborative mechanism breaks through the bottleneck of single-track resource supply, optimizes the rational allocation of computing and communication resources and spectrum resources, and improves the overall operational efficiency of large-scale satellite constellations. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0037] Figure 1. Schematic diagram of an energy-efficient routing system model in a low-Earth orbit satellite constellation;

[0038] Figure 2 is a schematic diagram of energy-saving routing optimization methods in low-Earth orbit satellite constellations. Detailed Implementation

[0039] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0040] As shown in Figures 1 and 2, in an embodiment of the present invention, an energy-saving routing method in a low-Earth orbit satellite constellation includes:

[0041] A system model is constructed that includes satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic. The core optimization objective is to minimize satellite battery life loss, and a joint optimization problem is established.

[0042] Based on the battery attenuation model and dynamic channel awareness, the joint optimization problem is modeled as a Markov decision process in the finite time domain. Its elements include a state space defined by the task queue, channel quality and track coordination state, an action space defined by transmission power, antenna gain and task operation switch, and a reward function that includes battery life penalty and QoS reward.

[0043] A two-layer optimization framework integrating Q-learning algorithm and back-induction battery management algorithm is adopted to solve the Markov decision process. The upper layer dynamically optimizes the communication resource configuration parameters through Q-learning, and the lower layer optimizes the task scheduling sequence through back-induction.

[0044] Based on the solution results, combined with cross-track link cost calculation and dynamic feedback mechanism, the final communication resource configuration and task scheduling strategy are output, and adaptive adjustments are made according to real-time battery status and channel quality.

[0045] In this embodiment, a model for energy-saving routing of a low-Earth orbit satellite constellation is constructed. The system model includes a satellite constellation topology model, an energy consumption model, an eclipse period determination model, and a mission traffic model. The satellite constellation topology model consists of an Iridium satellite constellation, which is composed of satellites and their inter-satellite links (ISLs). Intercontinental communication traffic is routed and transmitted through this constellation.

[0046] While ensuring that the QoS requirements of LEO satellite communication are met, satellite transmission parameters and routing strategies are optimized collaboratively to minimize the lifespan loss of satellite batteries, thereby extending the lifespan of satellite batteries and the overall constellation and reducing the operating cost of large-scale LEO satellite constellations.

[0047] Based on the established optimization problem, a multi-orbit LEO satellite energy-saving optimization method integrating reinforcement learning and Markov decision-making is proposed. Based on the battery decay model and dynamic channel perception, the system state space, action space and reward function are established to model the energy-saving optimization problem as a finite Markov decision process.

[0048] A two-layer optimization framework is designed by integrating Q-learning and reverse inductive battery management algorithm. The upper layer dynamically adjusts satellite transmission power and antenna gain through Q-learning, while the lower layer optimizes task acquisition-processing-transmission scheduling through reverse inductive battery management algorithm.

[0049] A cross-orbit cooperation mechanism is introduced to calculate the LEO-MEO / GEO link cost, correct the state transition probability and reward function, and obtain the optimal communication resource configuration and task scheduling strategy through iterative solution. Based on battery deep discharge constraints and channel quality feedback, the optimization strategy is adjusted in real time to ensure that the satellite battery SoC is maintained within a safe range, achieving a dynamic balance between task QoS and satellite lifetime.

[0050] In this embodiment, the method achieves coordinated optimization of satellite battery life, mission QoS, and resource utilization by constructing a multi-track collaborative model, integrating Q-learning and reverse inductive battery management algorithms, and introducing dynamic channel awareness and battery attenuation modeling.

[0051] As a preferred embodiment of the present invention, the satellite energy consumption model quantifies the energy consumption of the satellite in each stage of transmission, reception, nominal operation and data processing, and defines the dynamic update equation of battery energy and the quantitative equation of battery life loss related to the depth of discharge.

[0052] A solar eclipse period determination model is established based on the spatial geometric relationship between the Earth, the Sun and the satellite, which is used to identify in real time whether the satellite is in the sunshine period, the umbra or the penumbra.

[0053] A task traffic model based on geographical region division is established to generate a traffic matrix that varies with space and time. The strength of the traffic matrix is ​​positively correlated with the regional user density and the number of terminals, and negatively correlated with the distance between regions.

[0054] In this embodiment, the energy-saving routing model for a low-Earth orbit (LEO) satellite constellation consists of 66 LEO satellites and inter-satellite links. The satellites are evenly distributed across six polar orbital planes, with 11 satellites in each plane. The orbital altitude is set at 780 km, enabling seamless global coverage. Intercontinental communication traffic is routed and transmitted across regions and exchanged data through this constellation. Each satellite in the constellation is equipped with four inter-satellite links: two intra-orbit links connect adjacent satellites in the same polar orbit, remaining constant throughout satellite operation; the other two are inter-orbit links connect satellites in different polar orbits. Because the satellites are constantly in motion, the inter-orbit links exhibit dynamic characteristics.

[0055] The satellite's energy comes from solar panel charging and battery storage, employing a cyclical pattern of "charging during sunshine and discharging during eclipses." During sunshine periods, the solar panels capture solar energy to power the satellite's operation, and excess energy charges the batteries. During eclipses, with no solar input, the satellite relies entirely on battery storage to maintain operation and communication missions. The satellite's total energy consumption is defined as the sum of transmission energy consumption, reception energy consumption, nominal operational energy consumption, and data processing energy consumption. .

[0056] The battery energy dynamic update model is C h (t) represents the solar charging power, and τ represents the time slice length. Let t represent the total energy consumption during time period t. The quantitative model for battery life loss is directly related to the depth of discharge, which refers to the percentage of total battery capacity discharged during a single discharge cycle, defined as... , For the maximum battery capacity, C B (t) represents the remaining battery charge at time t. The battery lifespan depletion rate is determined by... Quantization, defined as Where D is the depth of discharge at the current moment, and A is a constant related to battery characteristics, which is determined by the battery's material and manufacturing process.

[0057] The eclipse period determination model is based on the spatial geometric relationship between the Earth, the Sun, and the satellite. It uses a conical shadow model to identify in real time whether the LEO satellite is in a shadow region with no or diminished solar energy input. The height of the eclipse cone is defined as... The height of the penumbra cone is defined as follows: The corresponding half vertex angle is defined as , D E D is the diameter of the Earth. S The diameter of the sun, The distance between the Earth and the Sun is given. The angle θ between the Sun, Earth, and satellite is calculated using real-time satellite orbit parameters. Determined to completely obscure the umbra, when Determined as partially obscured penumbra area, when It is determined to be a period of sunshine.

[0058] The core logic of the traffic model is that traffic demand is positively correlated with user density and number of terminals in a geographical region, and negatively correlated with the distance between regions. Based on this, a realistic traffic distribution that varies with time and space is generated. To quantify the traffic generation capacity of different regions globally, the paper divides the Earth into 288 geographical regions, i.e., a 12×24 grid, with each region having a latitude of 15° × longitude. Each region is assigned a user density level u. j Based on actual internet user distribution data, a higher density value indicates a denser user base in the region. Only 120 regions with significant traffic demand are retained: if a region has fewer than 1.6 million internet users annually, the density level is set to 0; otherwise, the density level is scaled proportionally. Terminal density level h j This is used to correlate "regional user density" with "total number of terminals" to quantify the traffic generation weight of a single region. The calculation formula is as follows:

[0059] ,in, Total number of terminals. Traffic matrix T ij The flow intensity between "source region i" and "destination region j" is defined and is the core output of the flow model. The calculation formula is as follows: , This represents the geographical distance between the center of source region i and the center of destination region j.

[0060] In a preferred embodiment of the present invention, the objective function of the joint optimization problem is:

[0061] ;

[0062] in, and For weighting coefficients, Qos total To normalize the overall QoS index, This represents the average battery life consumption of all satellites during the time period from t1 to t2.

[0063] In this embodiment, system parameters are initialized, including satellite orbit parameters, battery parameters, and mission parameters, wherein the battery parameter is the maximum capacity EB. max Safety threshold EB min Attenuation coefficient A and temperature coefficient k T The task parameters include the data size B. s and task priority w p To address the limitation of satellite networks relying solely on solar energy during eclipses, three types of issues are considered: First, the issue of extending battery life and controlling depth of discharge (DOD). The number of charge-discharge cycles and the depth of discharge (DOD) of satellite batteries directly determine their lifespan, and existing routing does not consider battery status, leading to accelerated degradation. Second, the trade-off between energy consumption and service QoS. Satellite energy consumption involves transmission power, antenna gain, data processing / compression, and other aspects. Energy-saving operations may affect service quality, requiring the maintenance of QoS requirements such as latency, throughput, and data transmission quality (e.g., compression ratio control) while reducing energy consumption. Third, the routing adaptation issue in dynamic scenarios. The orbital motion of LEO satellites causes dynamic changes in network topology, link connections, and eclipse status. It is necessary to cope with random solar energy input and time-varying service traffic to avoid energy waste or QoS degradation caused by delayed routing decisions.

[0064] In energy-constrained LEO satellite network scenarios, an optimization objective function is constructed by jointly considering minimizing battery consumption and maximizing QoS service quality.

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] The constraints include energy and battery constraints, and service and QoS constraints. C1 represents the remaining battery power constraint. C1 represents the maximum permissible depth of discharge; C2 represents the depth of discharge constraint, where 0.8 is the safe depth of discharge threshold; C3 represents the transmission power constraint; and C4 represents the time delay constraint, PT. x，y C5 represents the propagation delay of the link (x,y); C5 represents the throughput constraint; T represents the propagation delay of the link (x,y); C5 represents the propagation delay of the link (x,y); and T represents the propagation delay of the link (x,y). reqFor business demand rate.

[0071] in, ; Determined by the depth of discharge. Battery life consumption. The definition is as follows:

[0072] ;

[0073] Qos total The normalized comprehensive QoS index, which integrates latency, throughput, and transmission quality, is defined as follows:

[0074] ;

[0075] in , and These are the weights for latency, throughput, and transmission quality, respectively, satisfying... , To average routing delay, For the maximum allowable delay, Average throughput, For the theoretical maximum throughput, E is the average data compression normalized mean square error. pmax This is the maximum permissible distortion threshold.

[0076] In a preferred embodiment of the present invention, the Q-value update of the upper-layer Q-learning algorithm in the two-layer optimization framework follows the following rules:

[0077] ;

[0078] Where α is the learning rate, γ is the discount factor, and the Q-table is initialized as follows: .

[0079] In this embodiment, the state space Where Q is the task queue length, Ch is the channel quality index (signal-to-noise ratio SNR, fading coefficient h), and Orbit is the orbit coordination state (0 for single LEO, 1 for LEO-MEO, 2 for LEO-GEO).

[0080] Action space , P is the transmission power. G is the antenna gain. For task acquisition switch, For task compression switch, Switch for task transmission;

[0081] State transition probability:

[0082] ;

[0083] in:

[0084] ; ;

[0085] reward function λ is the lifespan penalty coefficient. To process decision vectors.

[0086] Upper-level Q-learning optimization, Q-table initialized as The updated formula is:

[0087] , For learning rate, Let be the discount factor, and the convergence condition be... , This is the convergence threshold;

[0088] In a preferred embodiment of the present invention, the optimal action sequence is obtained by back-induction optimization of the task scheduling sequence at the lower level, based on the value function of the Bellman equation, and by back-iteration.

[0089] In this embodiment, the lower-level reverse inductive battery management algorithm is based on the derivation of the value function of the Bellman equation:

[0090] The optimal action sequence is solved by reverse iteration:

[0091] .

[0092] Q-learning output , As a constraint on the reverse inductive battery management algorithm, , Reverse induction of battery management output , , Feedback is sent to Q-learning to correct the QoS term in the reward function.

[0093] In a preferred embodiment of the present invention, the cross-track link cost calculation formula is as follows:

[0094] ;

[0095] in, Let x be the difference in orbital altitude between the satellite and y be the ... For propagation loss, Here, λ is the orbital weighting coefficient, and λ is the transmission wavelength. L p Losses due to rainfall.

[0096] In this embodiment, wherein, , ;

[0097] The dynamic energy consumption model is derived as follows:

[0098] ;

[0099] Where P dc To collect power, P bl P is the background load power. tr For transmission power, τ represents the duration of the communication link, and τ represents the time slice length.

[0100] In a preferred embodiment of the present invention, the dynamic feedback mechanism includes:

[0101] Battery safety dynamic constraints adjust the maximum permissible depth of discharge in real time based on the battery health status.

[0102] Channel adaptation constraint: When the real-time channel quality is lower than a preset threshold, it triggers compensation adjustment of antenna gain or transmission power.

[0103] Dynamic QoS constraints for tasks: The minimum transmission rate or maximum latency requirement is dynamically set based on task priority.

[0104] In this embodiment, the real-time adjustment strategy satisfies the following dynamic constraints and feedback mechanisms:

[0105] Battery safety constraints are dynamically updated. , , As the reference maximum depth of discharge, For adjustment purposes;

[0106] Channel adaptation constraints , , Noise ratio threshold, when At that time, antenna gain adjustment is triggered. ;

[0107] Task QoS constraints , This is the lower limit of the task transmission rate, which is dynamically adjusted according to the task priority (0.9 for urgent tasks and 0.7 for low priority tasks).

[0108] The LEO satellite energy-saving optimization system, based on single-orbit communication resource control, introduces three core technology modes: multi-orbit collaboration, dual-layer optimization algorithm, and dynamic channel sensing and battery attenuation modeling. These three technology modes complement each other and work synergistically to improve the service quality and energy-saving effect of the satellite network. The multi-orbit collaboration mode can also break through the bottleneck of single-orbit resources by effectively expanding the network coverage and resource supply capacity through the link cooperation between LEO and MEO / GEO satellites, adapting to the needs of large-scale satellite communication.

[0109] By establishing a multi-objective joint optimization problem involving mission QoS, satellite battery life, and resource utilization, the system state and action space are quantitatively defined. A finite-time-domain Markov decision process model is constructed, and a two-level iterative algorithm combining Q-learning and back induction is used to solve the problem until convergence or the maximum number of iterations is reached. This yields a preliminary scheme for communication resource allocation and mission scheduling. Subsequently, by combining a dynamic constraint feedback mechanism and real-time sensing of channel quality and battery status, the results are optimized.

[0110] On the one hand, it can make full use of multi-orbit resources and computing and communication resources to meet the QoS requirements of different priority tasks. On the other hand, it can also effectively reduce the deep discharge loss and dynamic energy consumption range of satellite batteries, extend the service life of satellites, and improve the overall operational stability and economy of large-scale LEO satellite constellations.

[0111] This invention also provides an energy-saving routing system in a low-Earth orbit satellite constellation, the system comprising:

[0112] The optimization module is used to construct a system model that includes satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic, with the core optimization objective of minimizing satellite battery life loss, and to establish a joint optimization problem.

[0113] The modeling module is used to model the joint optimization problem as a finite-time Markov decision process based on the battery attenuation model and dynamic channel awareness. Its elements include a state space defined by the task queue, channel quality and track coordination state, an action space defined by transmission power, antenna gain and task operation switch, and a reward function that includes battery life penalty and QoS reward.

[0114] The solution module is used to solve the Markov decision process using a two-layer optimization framework that integrates the Q-learning algorithm and the back-inductive battery management algorithm. The upper layer dynamically optimizes the communication resource configuration parameters through Q-learning, and the lower layer optimizes the task scheduling sequence through back-induction.

[0115] The strategy output module is used to output the final communication resource configuration and task scheduling strategy based on the solution results, combined with cross-track link cost calculation and dynamic feedback mechanism, and to adaptively adjust it according to real-time battery status and channel quality.

[0116] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An energy-saving routing method in a low-Earth orbit satellite constellation, characterized in that, The method includes: constructing a system model encompassing satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic, with minimizing satellite battery life loss as the core optimization objective, and establishing a joint optimization problem; based on a battery attenuation model and dynamic channel awareness, modeling the joint optimization problem as a finite-time Markov decision process, whose elements include a state space defined by mission queues, channel quality, and orbital coordination state; an action space defined by transmission power, antenna gain, and mission operation switches; and a reward function including battery life penalty and QoS reward; solving the Markov decision process using a two-layer optimization framework integrating Q-learning and back-inductive battery management algorithms, where the upper layer dynamically optimizes communication resource configuration parameters through Q-learning, and the lower layer optimizes the mission scheduling sequence through back-induction; based on the solution results, combined with cross-orbit link cost calculation and a dynamic feedback mechanism, outputting the final communication resource configuration and mission scheduling strategy, and adaptively adjusting it according to real-time battery status and channel quality.

2. The energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1, characterized in that, The satellite energy consumption model quantifies the energy consumption of the satellite in each stage of transmission, reception, nominal operation, and data processing, and defines the dynamic update equation of battery energy and the quantitative equation of battery life loss related to the depth of discharge. Based on the spatial geometric relationship between the Earth, the Sun, and the satellite, an eclipse period determination model is established to identify in real time whether the satellite is in the sunshine period, umbra, or penumbra. A task traffic model based on geographical region division is established to generate a traffic matrix that varies with space and time. The intensity of the traffic matrix is ​​positively correlated with the regional user density and the number of terminals, and negatively correlated with the distance between regions.

3. The energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1, characterized in that, The objective function of the joint optimization problem is: ;in, and For weighting coefficients, Qos total To normalize the overall QoS index, This represents the average battery life consumption of all satellites during the time period from t1 to t2.

4. The energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1, characterized in that, In the aforementioned two-layer optimization framework, the Q-value update of the upper-layer Q-learning algorithm follows the following rules: Where α is the learning rate, γ is the discount factor, and the Q-table is initialized to... 。 5. An energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1 or 4, characterized in that, In the lower layer, the task scheduling sequence is optimized through back induction. The value function based on the Bellman equation is derived, and the optimal action sequence is solved through back induction iteration.

6. The energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1, characterized in that, The formula for calculating the cost of the inter-railway link is as follows: ;in, Let x be the difference in orbital altitude between the satellite and y be the ... For propagation loss, Here, λ is the orbital weighting coefficient, and λ is the transmission wavelength. L p Losses due to rainfall.

7. The energy-saving routing method in a low-Earth orbit satellite constellation according to claim 1, characterized in that, The dynamic feedback mechanism includes: dynamic battery safety constraints, which adjust the maximum allowable depth of discharge in real time according to the battery health status; channel adaptation constraints, which trigger compensation adjustments to antenna gain or transmission power when the real-time channel quality is lower than a preset threshold; and dynamic task QoS constraints, which dynamically set the minimum transmission rate or maximum latency requirements according to task priority.

8. An energy-saving routing system in a low-Earth orbit satellite constellation, used to implement the energy-saving routing method in a low-Earth orbit satellite constellation as described in any one of claims 1-7, characterized in that, The system comprises: an optimization module for constructing a system model including satellite constellation topology, satellite energy consumption, eclipse period determination, and mission traffic, with minimizing satellite battery life loss as the core optimization objective, and establishing a joint optimization problem; a modeling module for modeling the joint optimization problem as a finite-time Markov decision process based on a battery attenuation model and dynamic channel awareness, the elements of which include a state space defined by mission queues, channel quality, and orbital coordination state, an action space defined by transmission power, antenna gain, and mission operation switches, and a reward function including battery life penalty and QoS reward; a solution module for solving the Markov decision process using a two-layer optimization framework that integrates Q-learning algorithm and back-inductive battery management algorithm, wherein the upper layer dynamically optimizes communication resource configuration parameters through Q-learning, and the lower layer optimizes the mission scheduling sequence through back-induction; and a policy output module for outputting the final communication resource configuration and mission scheduling policy based on the solution results, combined with cross-orbit link cost calculation and dynamic feedback mechanism, and adaptively adjusting according to real-time battery status and channel quality.

Images