Efficiency adaptation method applied to stratosphere solar communication unmanned aerial vehicle

Abstract

According to the efficiency adaptation method applied to the solar communication unmanned aerial vehicle in the stratosphere, collaborative optimization of energy management and communication service of the solar unmanned aerial vehicle in the stratosphere environment is achieved by constructing a dynamic solar radiation database, predicting communication load and setting a dynamic threshold value and a multi-objective optimization mechanism. The method has the capability of responding to solar radiation fluctuation and communication load change in real time, effectively avoids the problems of misalignment of electric quantity management and interruption of a communication link, and remarkably improves the cruising ability and task reliability of the unmanned aerial vehicle. Meanwhile, according to the method, through a data-driven closed-loop optimization mechanism, the energy utilization efficiency and the communication quality are both considered, and the method has good practicability and popularization value.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular to an efficiency adaptation method applied to a stratospheric solar communication unmanned aerial vehicle. BACKGROUND

[0002] With the continuous upgrading of global communication demand towards wide coverage, high reliability and low latency, stratospheric solar communication unmanned aerial vehicles gradually become a research hotspot in the field of communication technology due to their long endurance, flexible deployment and wide coverage, and are widely used in emergency rescue, remote area communication, ocean monitoring and other scenarios. The stratospheric unmanned aerial vehicle usually flies in the atmosphere at an altitude of 18-25 kilometers. The atmosphere in this area is thin, the wind speed is stable, and the sunlight is sufficient, which is conducive to the long-term hovering and communication service of the solar unmanned aerial vehicle. The existing stratospheric communication unmanned aerial vehicle relies on fixed meteorological model or static solar radiation database for path planning and energy management. Its system structure usually includes a solar panel, an energy storage device, a communication load and a flight control system, which executes tasks through a preset flight path and a static energy allocation strategy.

[0003] However, the stratosphere has particularity and complexity, which leads to multiple bottlenecks in the existing technology. On the one hand, the stratospheric solar radiation intensity is affected by factors such as latitude, season, solar elevation angle and ozone concentration, and there is significant short-term fluctuation. The fixed meteorological model or static radiation database used by the existing system cannot capture this fluctuation in real time, resulting in large power prediction errors and problems such as waste of overcharging or interruption of endurance due to undercharging. On the other hand, the communication load changes dynamically with the regional user density, and the existing technology does not establish a linkage mechanism between communication load and energy allocation. In high-load situations, communication links may be interrupted due to insufficient base station power supply, and in low-load situations, energy may be wasted due to redundant base station power supply. In addition, the existing stratospheric solar radiation database has data missing areas, and lacks a scientific data updating and verification mechanism, further leading to redundant unmanned aerial vehicle flight path planning and deviation in photovoltaic module efficiency calculation, which seriously restricts the overall efficiency and task reliability of the stratospheric solar communication unmanned aerial vehicle.

[0004] Therefore, there is an urgent need in the art for an efficiency adaptation method that can respond to environmental changes in real time and cooperatively optimize energy management and communication services to improve the endurance and communication stability of stratospheric solar communication unmanned aerial vehicles in complex environments. SUMMARY

[0005] The purpose of the present application is to provide an efficiency adaptation method applied to a stratospheric solar communication unmanned aerial vehicle to solve the problems existing in the prior art.

[0006] To achieve the above purpose, the present application provides the following solutions:

[0007] The application provides an efficiency adaptation method applied to a stratosphere solar communication unmanned aerial vehicle, and comprises the following steps:

[0008] S1. A global dynamic stratosphere solar radiation intensity database is constructed, and the database comprises a satellite data layer, a real-time unmanned aerial vehicle layer and a fusion layer;

[0009] S2. Based on historical communication load data and real-time user access data, the communication load level in the next one hour is predicted through an LSTM and attention mechanism combined model, and the communication load level is divided into three levels of low, medium and high;

[0010] S3. The solar radiation data, communication load data and longitude, latitude and altitude information are corresponded one by one to generate a dynamic comprehensive geographic information map;

[0011] S4. A two-dimensional dynamic threshold is set, and the collected data and the database data are compared and verified, and if the threshold is exceeded, the database is updated through unmanned aerial vehicle cross verification and satellite re-measurement;

[0012] S5. According to the task priority, the weights of the energy target, the communication target and the efficiency target are dynamically allocated, and the flight path, the attitude and the power distribution strategy of the unmanned aerial vehicle are optimized in real time.

[0013] Preferably, in step S1, the specific steps of constructing the global dynamic stratosphere solar radiation intensity database comprise:

[0014] Global stratosphere solar radiation data is collected by a low-orbit satellite at a frequency of 15 minutes / time, and 3σ criterion is used for outlier rejection;

[0015] Local solar radiation, ozone concentration and temperature data are collected by a sensor carried by the unmanned aerial vehicle at a frequency of 2 minutes / time, and are encrypted and transmitted;

[0016] An adaptive Kalman filtering algorithm is used to fuse the satellite data and the unmanned aerial vehicle data to obtain accurate solar radiation intensity data, and the state prediction formula is:

[0017]

[0018] Wherein, is the predicted value of the solar radiation intensity at time k after fusion; A is a state transition matrix, and A=1 is taken; is the state estimation value at time k-1; B is a control input matrix, and B=0 is taken; u k-1 is the control input at time k-1, and u k-1 =0 is taken.

[0019] Preferably, the adaptive Kalman filtering algorithm further comprises a dynamic weight fusion step, comprising:

[0020] Computing satellite data variance and UAV data variance

[0021] According to the variance, a dynamic weight is calculated:

[0022]

[0023]

[0024] The observation value is updated:

[0025] z k = w s × z (s,k) + w d × z (d,k) ;

[0026] wherein z k is the observation value at time k; z (s,k) is the satellite observation data at time k; and z (d,k) is the UAV observation data at time k.

[0027] Preferably, in step S2, the LSTM combined with the attention mechanism model is a 2-layer LSTM combined with a time attention mechanism model, and the attention weight calculation formula thereof is:

[0028]

[0029] wherein, is the weighted cosine similarity.

[0030] Preferably, in step S3, the dynamic comprehensive geographic information map is generated by a MODTRAN 6.0 model and an ozone absorption correction model, and the ozone absorption correction formula is:

[0031] P c = P r × e (-α×C×L) ;

[0032] wherein P c is the corrected radiation intensity, a is the ozone absorption coefficient, C is the ozone concentration, and L is the radiation path length.

[0033] Preferably, in step S4, the two-dimensional dynamic threshold is dynamically adjusted according to latitude and season, and the basic threshold formula is:

[0034] Th(lat) = 900 × (1-0.006 × |lat|) × cos(90°-β);

[0035] wherein β is the solar elevation angle, and when β≤10°, the threshold is forcibly set to 300 W / m 2 .

[0036] Preferably, in step S5, the energy target is that the average radiation intensity of the flight line is not less than the historical dynamic setting value;

[0037] The communication target is that the signal-to-noise ratio SNR is greater than or equal to 15 dB, and the bit error rate BER is less than or equal to 10-5.

[0038] The efficiency target is that the total length of the flight line is not more than 120% of the minimum length of the task.

[0039] Preferably, the target optimization process adopts an intelligent optimization algorithm, and the parameter settings are as follows: population size 30, iteration number 12, crossover probability 0.7, and mutation probability 0.05.

[0040] Preferably, the communication load level is associated with the base station power supply strategy:

[0041] When the load is low, the base station power supply ratio is reduced to 30%;

[0042] When the load is medium, the base station power supply ratio is maintained at 50%;

[0043] When the load is high, the base station power supply ratio is increased to 70%, and the standby capacitor is enabled.

[0044] Preferably, it also includes encrypting the data collected by the unmanned aerial vehicle for encrypted transmission through an encryption algorithm, and the encryption formula is:

[0045] C=E k (M);

[0046] Wherein, M is the original real-time data collected by the solar unmanned aerial vehicle, k is the encryption key, E is the encryption algorithm, and C is the encrypted data.

[0047] The present application has the following beneficial technical effects compared with the prior art:

[0048] The efficiency adaptation method applied to the stratospheric solar communication unmanned aerial vehicle provided by the present application realizes the collaborative optimization of energy management and communication service of the solar unmanned aerial vehicle in the stratospheric environment through the construction of a dynamic solar radiation database, the prediction of communication load, the setting of a dynamic threshold value and a multi-target optimization mechanism. The method has the ability to respond to solar radiation fluctuations and communication load changes in real time, effectively avoids the problems of inaccurate power management and communication link interruption, and significantly improves the endurance and task reliability of the unmanned aerial vehicle. At the same time, the method takes into account the energy utilization efficiency and communication quality through a data-driven closed-loop optimization mechanism, and has good practicability and promotional value. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to make the technical solutions in the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description only constitute some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0050] Figure 1 The application provides an efficiency adaptation method applied to a stratosphere solar communication unmanned aerial vehicle.

[0051] Figure 2 The application provides a database updating flowchart in the efficiency adaptation method applied to the stratosphere solar communication unmanned aerial vehicle.

[0052] Figure 3 The application provides a database establishing flowchart in the efficiency adaptation method applied to the stratosphere solar communication unmanned aerial vehicle.

[0053] Figure 4 The application provides an unmanned aerial vehicle control flowchart in the efficiency adaptation method applied to the stratosphere solar communication unmanned aerial vehicle.

[0054] Figure 5 The application provides an application scenario schematic diagram of the efficiency adaptation method applied to the stratosphere solar communication unmanned aerial vehicle. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort belong to the protection scope of the present application.

[0056] The present application aims to provide an efficiency adaptation method applied to a stratosphere solar communication unmanned aerial vehicle to solve the problems in the prior art.

[0057] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Embodiment 1

[0059] The present embodiment provides an efficiency adaptation method applied to a stratosphere solar communication unmanned aerial vehicle, as shown in Figures 1-5 The present embodiment provides an efficiency adaptation method applied to a stratosphere solar communication unmanned aerial vehicle, as shown in

[0060] 1. Low-orbit satellite global stratosphere environment data collection

[0061] The global stratosphere region solar radiation raw data is collected by low-orbit satellites, and the data update frequency is set to every 15 minutes / time. After analyzing and converting the raw solar radiation data, the abnormal value detection method based on 3σ criterion is used to eliminate the abnormal values in the satellite data, as follows:

[0062] Abnormal value detection method: calculate the mean μ and standard deviation σ of the raw data. If a data point x satisfies |x-μ|>3σ, it is determined that the data is an abnormal value and is eliminated, i.e. x 异常 ={x∣|x-μ|>3σ}, where the mean of the same batch of satellite collected data is (n is the number of the same batch of data), and the standard deviation is

[0063] 2. Real-time collection of local stratosphere environmental data by solar-powered unmanned aerial vehicles

[0064] The stratosphere solar communication unmanned aerial vehicle carries solar radiation, ozone concentration and temperature sensors, etc., to collect local stratosphere solar radiation intensity, ozone concentration, photovoltaic module surface temperature, etc. in the flight area. The data update frequency is set to every 2 minutes / time, and the real-time data collected by the solar-powered unmanned aerial vehicle is encrypted using a specific encryption protocol. The encrypted data is transmitted to the distributed database through the communication link. The encryption formula is C=E k (M), where M is the original real-time data collected by the solar-powered unmanned aerial vehicle, k is the encryption key, E is the encryption algorithm, and C is the encrypted data. The data receiving end restores the data by decryption algorithm, and the decryption formula is M=D k (C), where D is the decryption algorithm.

[0065] 3. Establishing a distributed stratosphere solar radiation intensity database

[0066] The global stratosphere environmental data collected by low-orbit satellites and the local real-time stratosphere environmental data collected by solar-powered unmanned aerial vehicles are transmitted to the distributed database to establish a global stratosphere solar radiation intensity database containing satellite data layer, unmanned aerial vehicle real-time layer and fusion layer. Each layer is defined as follows:

[0067] Satellite data layer: the preprocessed low-orbit satellite global solar radiation data is stored according to the collection time, latitude and longitude, altitude, etc., and a spatio-temporal index structure is used to record the solar radiation historical data and real-time update data of the stratosphere at different locations around the world, with an update frequency of 15 minutes / time.

[0068] Unmanned aerial vehicle real-time layer: the decrypted local stratosphere real-time data of the unmanned aerial vehicle is stored in association according to the flight time, latitude and longitude, and altitude information, and records the high-frequency real-time environmental data of the local stratosphere, with an update frequency of 2 minutes / time.

[0069] Constructing a fusion layer: An adaptive Kalman filter algorithm is used to eliminate environmental radiation noise. Global stratospheric data from the satellite data layer and local stratospheric data from the UAV real-time layer are fused to obtain accurate solar radiation intensity data. The algorithm steps and formulas are as follows:

[0070] (1) State prediction: in is the predicted solar radiation intensity at time k after fusion; A is the state transition matrix (stratospheric radiation data changes smoothly in a short time, so A = 1 is taken); Here is the state estimate at time k-1; B is the control input matrix (if there is no external control input, B = 0); u k-1 The control input at time k-1 (take u) k-1 =0).

[0071] (2) Error covariance prediction: P k|k-1 =AP k-1|k-1 A T +Q k-1 , where P k|k-1 Let P be the prediction error covariance matrix at time k. k-1|k-1 Let Q be the error covariance matrix estimated at time k-1. k-1 The process noise covariance moment at time k-1 (initial value set based on historical data).

[0072] (3) Gain calculation: K k =P k|k-1 H T HP k|k-1 H T +R k ) -1 , where K k R is the Kalman gain at time k; H is the observation matrix (mapping the state variables to the observations, H = 1); k The observation noise covariance matrix at time k (pre-set based on the accuracy of satellite and UAV sensors).

[0073] (4) Variance of satellite data within a single acquisition period: (n d This represents the amount of valid data generated by the drone during this period. (This is the average value of drone data).

[0074] (5) Variance of solar-powered drone data: (n d This represents the amount of valid data generated by the drone during this period. (This is the average value of drone data).

[0075] (6) Dynamic weight formula: Observation update is z k = w s × z (s,k) + w d × z (d,k) , where z k is the observation at time k, z (s,k) is the satellite observation data at time k, and z (d,k) is the UAV observation data at time k. In addition, if the data missing rate of a data source is > 30%, its weight is automatically reduced within a certain range; if the data source returns to normal (i.e., the missing rate < 10%), the dynamic weight calculation is restored.

[0076] ⑺. State estimation update: where, is the final solar radiation intensity data at time k.

[0077] ⑻. Error covariance update: P k|k = (I - K k H)P k|k-1 , where I is the identity matrix.

[0078] ⑼. Adaptive adjustment of process noise covariance matrix: According to the observation residual , the residual covariance is calculated and compared with the theoretical residual covariance S k|k-1 = HP k|k-1 H T + R k , and the process noise covariance matrix Q k = a k Q k-1 is adjusted by the scaling factor (tr is the matrix trace operation), to realize the adaptive optimization of the algorithm.

[0079] 4. Predicting the communication load level:

[0080] Based on historical communication load data and real-time user access data collected by the communication UAV, the data set is divided into a training set (70% historical data), a validation set (20% historical data), and a test set (10% historical data + real-time collected data) in chronological order. The input sequence is constructed with a time step T = 12 (15 minutes per step), and a joint model of 2-layer LSTM and time attention mechanism is used to predict the future 1-hour communication load. According to the regional user density, the communication load level is divided into three levels: low, medium, and high, and is transmitted to the database, as follows:

[0081] 2-layer LSTM: 64 neurons per layer, tanh as hidden layer activation function, sigmoid as gate activation function, dropout layer (dropout rate = 0.2) is set in each layer to prevent overfitting. The first layer outputs a 12-step sequence The second layer is based on H 1 Output deep sequence (64 dimensions per step).

[0082] Time attention mechanism: after the LSTM layer output, the similarity between the 12 time steps is normalized by Softmax to get the attention weight After dynamic allocation and weighted summation of the 12 time step weights, a feature vector that integrates key information is obtained where, is the output of the LSTM at step t, is the mean of the 12-step output, is the weighted cosine similarity, where λ = 0.01.

[0083] Output layer: map H att to a three-dimensional vector z = W out ·H att + b out , and then use the Softmax activation function to output the probability of the grade , which respectively correspond to low / middle / high load.

[0084] Model training and optimization: optimizer is Adam, learning rate η = 0.001, β1 = 0.9, β2 = 0.999, ε = 10 -8 ; the loss function is cross-entropy loss (y k is the one-hot label of the true load grade); training rounds (Epoch) 50 rounds, using early stopping strategy (stop training if the validation set loss does not decrease for 5 consecutive rounds), batch size (Batch Size) 32.

[0085] Low load: user density < 50 users / km2, base station power supply ratio reduced to 30%, prefer solar-powered unmanned aerial vehicle power supply;

[0086] Medium load: user density 50-150 users / km2, base station power supply ratio maintained at 50%, maintain power balance;

[0087] High load: user density > 150 users / km2, base station power supply ratio increased to 70%, standby capacitor is enabled to support the base station independently, and the remaining capacity of the capacitor is monitored in real time, without affecting the main energy distribution.5, form a comprehensive stratospheric geographic information map:

[0088] Using the MODTRAN 6.0 model and an ozone absorption correction model, the ozone layer's absorption error for ultraviolet radiation is corrected. Solar radiation data and communication load from the database are mapped one-to-one with latitude, longitude, and altitude to generate a dynamic integrated geographic information map. Input data is synchronized from both satellite data and UAV real-time data layers. Furthermore, this information map can optimize photovoltaic module efficiency calculation parameters based on stratospheric atmospheric density characteristics; the specific formula is as follows:

[0089] MODTRAN 6.0 model: P r =P0×τ a ×τ o,s Where P0 is the standard value of solar radiation at the upper boundary of the atmosphere; τ a τ represents the transmittance of non-ozone components in the stratosphere. o,s Transmittance of the built-in standard ozone profile; P r The original radiation intensity (unit: W / m²).

[0090] Ozone absorption correction model: P c =P r ×e (-α×C×L) , where P c The corrected radiation intensity is given by α, where α is the ozone absorption coefficient at the corresponding wavelength, C is the ozone concentration at the current altitude, and L is the radiation transmission path length. 6. Solar-Powered Communication Drone: The drone navigates along the planned route.

[0091] The database plans flight routes for solar-powered communication drones based on the coverage area of ​​the communication task, prioritizing the avoidance of low-energy areas with weak radiation, so as to enable stratospheric solar-powered drones to complete communication tasks in areas with sufficient energy.

[0092] 7. Real-time data collection:

[0093] During the flight, the sensors on the solar-powered communication drone simultaneously collect data such as solar radiation intensity, photovoltaic module surface temperature, and ozone concentration, and transmit them to the database.

[0094] 8. Set thresholds and update data:

[0095] Based on the radiation fluctuation characteristics at different latitudes in the stratosphere, a two-dimensional dynamic threshold for latitude and season is set:

[0096] Basic threshold: Th(lat)=900×(1-0.006×|lat|)×cos(90°-β))(lat∈[-90°,90°]), where lat is latitude (°) and β is solar altitude angle. When β≤10°, that is, when the sun is close to the horizon and the radiation intensity is extremely low, the threshold Th(lat) is forcibly set to 300W / ㎡ to avoid meaningless threshold determination.

[0097] Final threshold: Th final = Th(lat) x K sea Where K sea is the seasonal correction coefficient based on the change of the sun angle. K sea Based on the 10-year history of stratospheric radiation data, the correlation between latitude, season and radiation fluctuation is generated by Gaussian regression model fitting.

[0098] If the difference between the collected data and the database exceeds the final threshold Th final , cross-validation with unmanned aerial vehicles in the same area is performed first, and then the communication satellite is requested to conduct spot re-measurement in the area. Then the re-measured data and the unmanned aerial vehicle data are weighted and averaged in a certain proportion. If it still exceeds the final threshold Th final , the database information is updated to ensure the continuity and accuracy of the data.

[0099] 9. Three-dimensional optimization of communication link quality:

[0100] Solar unmanned aerial vehicles collect signal-to-noise ratio (SNR), bit error rate (BER) and other data every second and transmit them to the database to realize real-time monitoring of communication link quality. Intelligent optimization algorithm (population size 30, iteration number 12, crossover probability 0.7, mutation probability 0.05) is adopted to optimize three targets based on the updated comprehensive geographic information map: energy target, communication target and efficiency target.

[0101] Energy target: The average radiation intensity target value of the flight route is dynamically set according to the historical data of the task area;

[0102] Communication target: Enable temporary power supply with backup battery to maintain SNR≥15dB, BER≤10 -5 , to ensure the reliability of communication;

[0103] Efficiency target: The total length of the flight route does not exceed 120% of the task requirement, i.e. (L p is the planned flight route length, L r is the minimum length of the task), to avoid redundant flight;

[0104] According to the task priority, the weights of the three targets are dynamically allocated to reasonably adjust the proportion of flight power consumption and communication base station power supply, and to optimize the flight attitude and route of the unmanned aerial vehicle. At the same time, the radiation data collected by the optimized flight path is used as a high-priority sample to trigger the database update process, realizing a closed loop of "optimization-collection-update-reoptimization", taking into account the stratospheric endurance of solar unmanned aerial vehicles and the communication task requirements.

[0105] The application breaks through the limitation of fixed weather model and static data through a dynamic database and a real-time verification mechanism, responds to solar radiation fluctuation and communication load change in real time, avoids problems such as power management inaccuracy and communication link interruption. At the same time, the three targets of energy, communication and efficiency are simultaneously optimized, the precise adaptation of flight and energy management is realized through dynamic weight distribution, and the endurance and communication reliability of the unmanned aerial vehicle are improved. In addition, the stratosphere solar radiation intensity database can also be used in the fields of weather measurement, aircraft navigation assistance and the like, and the practicability and popularization value of the technical scheme are improved.

[0106] The principle and implementation mode of the present application are described by applying specific examples, and the above examples are only used to help understand the method and core idea of the present application; meanwhile, for the general technical personnel in the field, the specific implementation mode and application range will be changed according to the idea of the present application. In conclusion, the content of the specification should not be understood as the limitation of the present application.

Claims

1. A performance adaptation method for stratospheric solar-powered communication drones, characterized in that, Includes the following steps: S1. Construct a global dynamic stratospheric solar radiation intensity database, which includes a satellite data layer, a UAV real-time layer, and a fusion layer; S2. Based on historical communication load data and real-time user access data, the communication load level for the next hour is predicted by a joint model of LSTM and attention mechanism, and it is divided into three levels: low, medium and high. S3. Correspond solar radiation data and communication load data with latitude, longitude and altitude information to generate a dynamic integrated geographic information map; S4. Set a two-dimensional dynamic threshold, compare and verify the collected data with the database data, and update the database after cross-verification by UAV and retesting by satellite if the threshold is exceeded. S5. Dynamically allocate the weights of energy, communication, and efficiency targets based on task priorities, and optimize the UAV's flight path, attitude, and power allocation strategy in real time.

2. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, In step S1, the specific steps for constructing the global dynamic stratospheric solar radiation intensity database include: Global stratospheric solar radiation data were collected via low-Earth orbit satellites at a frequency of 15 minutes per time, and outlier removal was performed using the 3σ criterion. The drone uses sensors to collect local solar radiation, ozone concentration, and temperature data every 2 minutes and transmits them in encrypted form. An adaptive Kalman filter algorithm is used to fuse satellite and UAV data to obtain accurate solar radiation intensity data. The state prediction formula is as follows: in, Let be the predicted solar radiation intensity at time k after fusion; A is the state transition matrix, and A = 1 is taken. Here is the state estimate at time k-1; B is the control input matrix, set to B=0; u k-1 Let u be the control input at time k-1. k-1 =0.

3. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 2, characterized in that, The adaptive Kalman filter algorithm further includes a dynamic weight fusion step, comprising: Calculate the variance of satellite data And drone data variance Calculate dynamic weights based on variance: Update observations: With k =in s ×z (s,k) +in d ×z (d,k) ; Among them, z k The value is the observation at time k; z (s,k) For time k, satellite observation data is used; for z (d,k) The data is from UAV observations at time k.

4. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, In step S2, the joint LSTM and attention mechanism model is a 2-layer LSTM and temporal attention mechanism joint model, and its attention weight calculation formula is as follows: in, This is the weighted cosine similarity.

5. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, In step S3, the dynamic integrated geographic information map is generated using the MODTRAN 6.0 model and the ozone absorption correction model. The ozone absorption correction formula is: P c =P r ×e (-α×C×L) ; Among them, P c The corrected radiation intensity is given by α, where α is the ozone absorption coefficient, C is the ozone concentration, and L is the radiation path length.

6. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, In step S4, the two-dimensional dynamic threshold is dynamically adjusted according to latitude and season, and its basic threshold formula is: Th(lat)=900×(1-0.006×|lat|)×cos(90°-β); Where β is the solar altitude angle, and when β≤10°, the threshold is forcibly set to 300W / m. 2 .

7. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, In step S5, the energy target is that the average radiation intensity of the flight path is not lower than the historical dynamic set value; The communication objectives are a signal-to-noise ratio (SNR) ≥ 15 dB and a bit error rate (BER) ≤ 10⁻⁵. The efficiency target is that the total length of the route should not exceed 120% of the minimum mission length.

8. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 7, characterized in that, The target optimization process uses an intelligent optimization algorithm with the following parameters: population size 30, number of iterations 12, crossover probability 0.7, and mutation probability 0.

05.

9. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, Communication load levels are related to base station power supply strategies: Under low load, the base station's power supply share drops to 30%; Under medium load conditions, the base station's power supply ratio remains at 50%. Under high load, the base station power supply ratio increases to 70%, and backup capacitors are activated.

10. The performance adaptation method for stratospheric solar-powered communication UAVs according to claim 1, characterized in that, This also includes encrypting the data collected by the drone during transmission using an encryption algorithm. The encryption formula is as follows: C=E k (M); Where M represents the raw real-time data collected by the solar-powered drone, k is the encryption key, E is the encryption algorithm, and C is the encrypted data.

Images