A method for optimizing dynamic channel transmission characteristics of a boundary layer meteorological observation aircraft

Abstract

The application discloses a boundary layer meteorological observation aircraft dynamic channel transmission characteristic optimization method. In the air-to-ground communication in the boundary layer in which a meteorological observation aircraft participates, assuming that a direct path between a transmitting end and a receiving end is blocked by an obstacle, part of signals emitted by the transmitting end is reflected to the receiving end through moving scatterers, and part of the signals directly reaches the receiving end without interference. The application describes a communication environment between the moving transmitting end and the moving receiving end by adopting a geometric random modeling mode, establishes a DRL-enabled boundary layer meteorological observation aircraft channel transmission characteristic dynamic optimization method based on geometric characteristics, and discloses a nonlinear relationship between the motion characteristics of the meteorological observation aircraft and a channel model. The application has important application value for evaluating a DRL-aided air-to-ground wireless communication system performance in the boundary layer.

Description

Technical Field

[0001] This invention relates to a method for optimizing the dynamic channel transmission characteristics of a boundary layer meteorological observation aircraft, belonging to the interdisciplinary field of atmospheric sounding and wireless communication. Background Technology

[0002] In recent years, with the development of meteorological observation technology towards intelligence and three-dimensionality, boundary layer meteorological observation unmanned aerial vehicles (UAVs) have become an important low-altitude detection platform, and the reliability of their data transmission directly affects the accuracy of atmospheric boundary layer research.

[0003] Modern meteorological observation networks need to meet the dual requirements of scientific research and industrial applications:

[0004] (1) Scientific observation dimension: By integrating satellite remote sensing, UAV detection and ground observation station data, boundary layer UAVs have become the core carrier for obtaining key meteorological parameters such as boundary layer turbulence, temperature and humidity profiles, thanks to their advantages such as high maneuverability, high vertical detection resolution and ability to penetrate clouds.

[0005] (2) Economic application dimension: The low-altitude meteorological data resource system constructed by the meteorological bureau requires that the UAV platform not only collect data, but also support economic service functions such as dynamic control of flight routes, which puts forward a higher standard for the real-time data transmission.

[0006] However, under complex atmospheric conditions (such as strong convection and wind shear), the aircraft's communication link faces a dual challenge:

[0007] (1) Scientific level: The severe signal attenuation and rapid time-varying channel directly affect the integrity of high-frequency data such as turbulence spectrum, leading to an increase in the error of boundary layer turbulence energy cascade analysis;

[0008] (2) Economic aspects: The Shenzhen drone logistics test showed that communication interruption during micro-burst weather can delay the replanning of delivery routes by up to 12 seconds and increase the operating cost per flight by 23%.

[0009] Deep Reinforcement Learning (DRL), originally developed as an intelligent algorithm to effectively address the non-convexity and high-dimensional computational complexity of complex optimization problems, has been increasingly adopted in recent years for various UAV wireless communication optimization problems. However, existing channel models remain insufficient in handling high dynamism, environmental complexity, and real-time computational requirements. They struggle to guarantee the timeliness and robustness of system performance optimization in complex air-to-ground environments, especially in boundary layer sudden weather scenarios. Furthermore, they fail to meet the dynamic service demands of the low-altitude economy. Therefore, developing dynamic optimization methods that simultaneously satisfy scientific accuracy and economic efficiency has become crucial for overcoming the bottlenecks in the coordinated development of smart meteorological observation and the low-altitude economy. Summary of the Invention

[0010] This invention provides a method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft, which solves the problems disclosed in the background art.

[0011] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0012] A method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft:

[0013] The non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains are analyzed in a pre-constructed DRL-based boundary layer meteorological observation aircraft channel model.

[0014] In the pre-constructed boundary layer meteorological observation aircraft channel model based on DRL, the non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains are combined with the flight energy consumption of the meteorological observation aircraft and the achievable rate in the channel. The energy efficiency optimization objective and constraints of the meteorological observation aircraft in the channel are constructed, and the reward function of the meteorological observation aircraft is defined.

[0015] Under constraints, the energy efficiency optimization objective is solved, and the solution is used to plan the optimal flight path based on the reward function of the meteorological observation aircraft to obtain the optimal action parameters of the meteorological observation aircraft adapted to non-stationary channels.

[0016] The method for constructing the boundary layer meteorological observation aircraft channel model based on DRL includes: considering that the scattering cluster is mobile, when the signal emitted by the transmitter reaches the receiver through the reflection path of the scattering cluster, constructing functional expressions for the transmission distance between the transmitter and receiver and the scattering cluster; and constructing a channel complex impulse response functional expression for the transmission path of the signal reflected by the scattering cluster to reach the receiver.

[0017] When the signal emitted by the transmitter reaches the receiver via a direct path, construct a functional expression for the transmission distance between the transmitter and receiver; construct a channel complex impulse response functional expression for the transmission path from the transmitter to the receiver.

[0018] Furthermore, the method for constructing a boundary layer meteorological observation vehicle channel model based on DRL is as follows: Establish a three-dimensional space... In a rectangular coordinate system, the line connecting the midpoint of the transmitting antenna array projection and the midpoint of the receiving antenna array is defined as follows: Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The axis is Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The lines of the plane are Axis; When a signal emitted by a transmitter reaches the receiver after multiple reflections, we model scattering as the first reflection after the signal is emitted by the transmitter, and the last reflection before the signal reaches the receiver, respectively by... and It consists of several scatterers, and the complex multipath between the first and last reflections is characterized by a virtual link; the meteorological observation vehicle flies within the boundary layer region, with the horizontal plane constrained by... arrive and arrive Height limit is arrive ; and They are respectively The minimum and maximum coordinates of the axis; and They are respectively The minimum and maximum coordinates of the axis; and They are respectively The minimum and maximum coordinates of the axis. Since the DRL algorithm runs in discrete time steps, we divide the total running time into... A discrete segment, wherein the discrete time interval needs to be small enough to satisfy the continuous-time channel state.

[0019] Furthermore, when the signal emitted by the transmitter reaches the receiver via the reflection path of the scattering cluster, the method for constructing the functional expression of the transmission distance between the transmitter, receiver, and scattering cluster is as follows: calculate the transmission distance at the transmitter... The root antenna to the first scattering cluster Time-varying transmission distance of each scatterer ;

[0020] in, , , It is a sufficiently small discrete time interval to accurately capture the time-varying characteristics of the channel. This represents the distance vector from the origin to the first scattering cluster. This represents the distance from the midpoint of the first scattering cluster to the... The offset vector of each scatterer This represents the aircraft coordinates calculated by the DRL algorithm. Indicates the distance from the midpoint of the transmitting antenna to the [missing information]. The distance vector of the root antenna is expressed as:

[0021] ;

[0022] The unit distance vector from the spacecraft to the first scattering cluster is represented as: ,

[0023] ;

[0024] in, and These represent the angles between the line connecting the transmitter and the first scatterer and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; and These represent the azimuth and elevation angles of the transmitting antenna array, respectively.

[0025] Calculate the receiver's first The root antenna and the last scattering cluster in the first Time-varying transport distance between scatterers ;

[0026] in, This indicates the distance from the midpoint of the last scattering cluster to the first... The offset vector of each scatterer From the midpoint of the transmitting antenna to the... The distance vector of the root antenna is expressed as:

[0027] ;

[0028] The unit distance vector from the last scattering cluster to the receiver is represented as: ,

[0029] ;

[0030] in, , , and These represent the angles between the line connecting the transmitter and the first scatterer in the last scattering cluster and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; This indicates the azimuth angle of the transmitting antenna array.

[0031] Furthermore, the expression for the channel complex impulse response function of the transmission path reaching the receiver after reflection from the scattering cluster is:

[0032] ;

[0033] in, Represents Rice factor; To represent a complex number; Represents independent and uniformly distributed random phases. Indicates wavelength. Indicates the carrier frequency. Indicates the rate at which light travels; This represents the propagation delay of the virtual link between two scattering clusters.

[0034] Furthermore, when the signal emitted by the transmitter reaches the receiver via a direct path, the method for constructing the functional expression of the transmission distance between the transmitter and receiver is as follows:

[0035] Calculate the transmitter's first The first antenna and receiver Transmission distance between antennas , ;

[0036] unit distance vector from transmitter to receiver .

[0037] Furthermore, the expression for the channel complex impulse response function of the transmission path where the signal directly reaches the receiver includes:

[0038] The expression for the channel complex impulse response function of the transmission path where the signal directly reaches the receiver is: ;

[0039] in, and These represent the angles between the line connecting the transmitter and receiver and the horizontal and vertical planes, respectively.

[0040] The expression for the total channel complex impulse response function is:

[0041] ;

[0042] in, Indicates path delay. This represents the impulse function.

[0043] Furthermore, the method for analyzing the non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains in a pre-constructed boundary layer meteorological observation aircraft channel model of DRL includes: calculating the first... The first antenna and receiver The cross-correlation function in the spatial domain for the transmission paths between the antennas is:

[0044] ;

[0045] in, Indicates discrete-time subscripts. The time interval representing the time-varying characteristics of the channel. Represents the expectation function, and This represents the normalized antenna spacing between the transmitter and receiver.

[0046] Let the above spatial expression in The time-domain autocorrelation function expression is obtained as follows:

[0047] ;

[0048] When the signal emitted by the transmitter reaches the receiver after being reflected by the scatterer and directly incident on the receiver, the transmitter's first... The first antenna and receiver The frequency domain correlation function expression for the transmission path between the root antennas is:

[0049] ;

[0050] in, Indicates frequency interval.

[0051] Furthermore, considering the flight energy consumption of meteorological observation aircraft and the achievable rate in the channel, the method for constructing the energy efficiency optimization objective and constraints of meteorological observation aircraft in the channel includes: considering the flight energy consumption of meteorological observation aircraft and achievable rate in the channel The expression is:

[0052]

[0053] ;

[0054] in, and These represent the constant blade profile power and induced power of the meteorological observation aircraft in hovering state, respectively. This represents the constant power required for ascent or descent; furthermore, Indicates the rotor blade tip velocity. This refers to the average rotor induced speed during hovering; parameter Indicates the fuselage drag ratio. Represents rotor solidity. and These are air density and rotor disk area, respectively; It is the horizontal velocity component of the meteorological observation aircraft;

[0055] The objective function for optimizing the energy efficiency of meteorological observation aircraft is: ;

[0056] The conditional constraints corresponding to the state and action spaces are expressed as follows:

[0057] ;

[0058] in, This is the initial position of the meteorological observation aircraft. It is a designated area where meteorological observation aircraft can fly. and These are the velocity and position vectors updated by the DRL for the weather vehicle. This is the maximum battery power consumption of a meteorological observation aircraft. This represents the range of acceleration vectors in the body coordinate system of the weather vehicle. The angular velocity of the weather vehicle is used to update the three-dimensional rotation angle.

[0059] Furthermore, the reward function is defined as:

[0060] ;

[0061] in, It is a reward symbol. It is a penalty for meteorological observation aircraft flying out of the designated area. , and These are three different weight values. It is the maximum achievable rate of the channel.

[0062] Furthermore, methods for solving the energy efficiency optimization objective under constraints include: using a dual-delay depth deterministic policy gradient algorithm to solve the energy efficiency optimization objective of the meteorological observation aircraft;

[0063] Initialize the actor network and the dual evaluator network, and randomly select elements containing... The network is updated in mini-batch using samples, and the update method that minimizes the loss is expressed as:

[0064] ;

[0065] in, It is a loss function. This represents the parameters of the first evaluator network. This represents the parameters of the second evaluator network. This indicates the total number of samples. This represents the environmental state observed by the agent. Indicates the agent's state The action to choose. and These are the value estimates of the output state-action in the evaluator network.

[0066] The expression for softly updating the target network is:

[0067] ;

[0068] in, This represents the value of the target commenter network. Indicates that the agent performs an action. The feedback obtained from the environment then guides the agent to learn better actions. This represents the discount factor, which controls the weight of future rewards. This represents the minimum value estimate obtained from the two-evaluator network. Indicates the next environmental state. This represents the target policy for generating the next state. This refers to the exploratory noise added to the actions, allowing the agent to actively explore the environment during training and avoid getting trapped in local optima. This represents the parameters in the two evaluator networks, which are updated with a delay to make the target more stable.

[0069] The actor network updates its formula according to the delay policy gradient algorithm as follows:

[0070] ;

[0071] in, This indicates gradient calculation. This represents the gradient of the network parameters with respect to the actor's policy. Indicates the strategic objective. It is the expectation function, which takes the expectation of the distribution of environmental state and agent action to avoid single sample noise. Indicates an action, Indicates state, This represents the gradient of the commenter network towards the action. This represents the gradient of the policy network with respect to its own parameters. This represents the network parameters for two evaluators, while This represents the parameters of the actor policy network.

[0072] The beneficial effects achieved by this invention are as follows: This invention adapts to various complex boundary layer air-to-ground scenarios by designing flexible DRL core elements, such as states, actions, rewards, and various channel parameters, to achieve dynamic optimization and channel non-stationarity analysis; it can effectively reveal the statistical characteristics of the model; it has the ability to handle high-dimensional continuous spaces and time-varying problems, and can cope with challenges such as environmental noise, partial observability, and drastic state fluctuations; it can reduce computer training and simulation time and alleviate the data processing burden. Attached Figure Description

[0073] Figure 1 : A schematic diagram of the boundary layer wireless communication channel model based on DRL proposed in this invention;

[0074] Figure 2: A comparative schematic diagram of the algorithms for optimizing the flight path of meteorological observation aircraft in three-dimensional space according to the present invention;

[0075] Figure 3 Schematic diagram illustrating the impact of different numbers of scattering clusters on the average energy efficiency of DRL-based boundary layer meteorological observation aircraft;

[0076] Figure 4 Schematic diagram illustrating the impact of different Rice factors on the airspace autocorrelation characteristics of the boundary layer meteorological observation aircraft channel based on DRL. Detailed Implementation

[0077] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0078] The invention proposes a method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft, comprising the following steps:

[0079] Step 1: Construct a boundary layer meteorological observation vehicle channel model based on DRL, such as... Figure 1 As shown, establish in three-dimensional space Using a Cartesian coordinate system, setting model parameters, discretizing continuous time, and dividing the total runtime into... There are discrete segments, where the discrete time intervals need to be small enough to satisfy the continuous-time channel state. The line connecting the midpoint of the transmitter antenna array projection and the midpoint of the receiver antenna array is defined as... Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The axis is Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The lines of the plane are Axis. Furthermore, the scattering bodies in the transmission environment between the transmitter and receiver are in a moving state. Considering the motion state and position vectors of the first and last bouncing scattering clusters, respectively, by... and The system consists of several scatterers, with a virtual link formed between two scattering clusters. The meteorological observation aircraft flies within a region of the boundary layer, with the horizontal plane as the constraint. arrive as well as arrive Height limit is arrive .

[0080] Step 2: In the DRL-based boundary layer meteorological observation vehicle channel, when the signal emitted by the transmitter reaches the receiver after being reflected by the scatterer, construct functional expressions for the transmission distance between the transmitter and receiver and the scatterer. The specific steps are as follows:

[0081] Step 201: In the DRL-based boundary layer meteorological observation vehicle channel, when the signal emitted by the transmitter reaches the receiver after being reflected by a scatterer, the scatterer is in motion. Calculate the first... root Antenna to the indivual The time-varying propagation distance of the scatterer is:

[0082] ;

[0083] in, This represents the distance vector from the origin to the first scattering cluster. This represents the distance from the midpoint of the first scattering cluster to the th scattering cluster. The offset vector of each scatterer The coordinates of the meteorological observation aircraft were calculated using the DRL algorithm. Indicates the distance from the midpoint of the transmitting antenna to the [missing information]. The distance vector of the root antenna is expressed as:

[0084] ;

[0085] The unit distance vector from the meteorological observation aircraft to the first scattering cluster is denoted as:

[0086] ;

[0087] in, and These represent the angles between the line connecting the transmitter and the first scatterer and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; and These represent the azimuth and elevation angles of the transmitting antenna array, respectively.

[0088] Step 202: In the DRL-based boundary layer meteorological observation aircraft channel, calculate the receiver's first... root Antenna and the indivual The time-varying transport distance between scatterers is:

[0089]

[0090] in, This represents the distance from the midpoint of the last scattering cluster to the th scattering cluster. The offset vector of each scatterer Indicates the distance from the midpoint of the transmitting antenna to the [missing information]. The distance vector of the root antenna is expressed as:

[0091] ;

[0092] The unit distance vector from the last scattering cluster to the receiver is denoted as:

[0093] ;

[0094] in, and These represent the angles between the line connecting the transmitter and the first scatterer and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; This indicates the azimuth angle of the transmitting antenna array.

[0095] Step 3: Construct the channel complex impulse response function expression for the transmission path of the signal after reflection from the scatterer to the receiver:

[0096] ;

[0097] in, Represents Rice factor; To represent a complex number; Represents independent and uniformly distributed random phases. Indicates wavelength. Indicates the carrier frequency. Indicates the rate at which light travels; This represents the propagation delay of the virtual link between two scattering clusters.

[0098] Step 4: When the signal emitted by the transmitter directly reaches the receiver, calculate the functional expression for the transmission distance between the transmitter and receiver. The specific steps are as follows:

[0099] Step 401: In the DRL-based boundary layer meteorological observation vehicle channel, when the signal emitted by the transmitter directly reaches the receiver, calculate the transmitter's... The first antenna and receiver The transmission distance between the antennas is:

[0100] ;

[0101] Step 402: In the DRL-based boundary layer meteorological observation vehicle channel, the unit distance vector from the transmitter to the receiver is:

[0102] ;

[0103] Step 5: The expression for the channel complex impulse response function of the transmission path from the direct signal to the receiver is as follows:

[0104] ;

[0105] in, and These represent the angles between the line connecting the transmitter and receiver and the horizontal and vertical planes, respectively.

[0106] The expression for the total channel complex impulse response function is:

[0107]

[0108] Step 6: Analyze the non-stationary characteristics of the transmission path in different time, spatial, and frequency domains. The specific steps are as follows:

[0109] Step 601: In the DRL-based boundary layer meteorological observation vehicle channel, when the signal emitted by the transmitter reaches the receiver after being reflected by the scatterer and directly incident on the receiver, calculate the transmitter's first... The first antenna and receiver The cross-correlation function in the spatial domain for the transmission paths between the antennas is:

[0110] ;

[0111] in, Indicates discrete-time subscripts. The time interval representing the time-varying characteristics of the channel. Represents the expectation function, and This represents the normalized antenna spacing between the transmitter and receiver.

[0112] Step 602: In the air-to-ground wireless communication channel within the DRL auxiliary boundary layer, let the above spatial domain expression... The time-domain autocorrelation function expression can then be obtained as follows:

[0113] ;

[0114] Step 603: In the air-to-ground wireless communication channel within the DRL-assisted boundary layer, when the signal emitted by the transmitter reaches the receiver after being reflected by a scatterer and directly incident on it, calculate the transmitter's... The first antenna and receiver The frequency domain correlation function expression for the transmission path between the root antennas is:

[0115] ;

[0116] in, Indicates frequency interval.

[0117] Step 7: Analyze the energy efficiency optimization objectives of the meteorological observation aircraft. The specific steps are as follows:

[0118] Step 701: In the DRL-based boundary layer meteorological observation vehicle channel, the expressions for the flight energy consumption and reachability rate of the meteorological observation vehicle are considered as follows:

[0119]

[0120] ;

[0121] Step 702: In the DRL-based boundary layer meteorological observation vehicle channel, the energy efficiency optimization objective function and conditional constraints of the meteorological observation vehicle are expressed as follows:

[0122] ;

[0123] in, This is the initial position of the meteorological observation aircraft. It is a designated area where meteorological observation aircraft can fly. and These are the velocity and position vectors updated by the DRL for the weather vehicle. This is the maximum battery power consumption of a meteorological observation aircraft. This represents the range of acceleration vectors in the body coordinate system of the weather vehicle. The angular velocity of the weather vehicle is used to update the three-dimensional rotation angle.

[0124] Step 703: In the DRL-based boundary layer meteorological observation vehicle channel, a Markov decision process is used to describe the optimization process of the meteorological observation vehicle. The state space represents the coordinates and velocity vectors of the meteorological observation vehicle, and the action space represents the acceleration and three-dimensional rotation angle of the meteorological observation vehicle in the body coordinate system. The reward function is defined as:

[0125] ;

[0126] in, It is a reward symbol, used By combining the objective equation, we can evaluate the action and achieve trajectory optimization. It is a penalty for meteorological observation aircraft flying out of the designated area. , and These are three different weight values. It is the maximum achievable rate of the channel.

[0127] Step 8: Solve the energy efficiency optimization problem of the meteorological observation aircraft using the dual-delay depth deterministic strategy gradient algorithm. The specific steps are as follows:

[0128] Step 801: Initialize the actor network and the dual evaluator network, and randomly select... The update method that updates the network using mini-batch updates of samples and minimizes the loss can be expressed as:

[0129] ;

[0130] in, It is a loss function. This represents the parameters of the first evaluator network. This represents the parameters of the second evaluator network. This indicates the total number of samples. This represents the environmental state observed by the agent. Indicates the agent's state The action to choose. and These are the value estimates of the output state-action in the evaluator network.

[0131] Step 802: Soft update the target network, the expression is:

[0132] ;

[0133] in, This represents the value of the target commenter network. Indicates that the agent performs an action. The feedback obtained from the environment then guides the agent to learn better actions. This represents the discount factor, which controls the weight of future rewards. This represents the minimum value estimate obtained from the two-evaluator network. Indicates the next environmental state. This represents the target policy for generating the next state. This refers to the exploratory noise added to the actions, allowing the agent to actively explore the environment during training and avoid getting trapped in local optima. This represents the parameters in the two evaluator networks, which are updated with a delay to make the target more stable.

[0134] Step 803: The actor network is updated according to the delay policy gradient algorithm as follows:

[0135] .

[0136] in, This indicates gradient calculation. This represents the gradient of the network parameters with respect to the actor's policy. Indicates the strategic objective. It is the expectation function, which takes the expectation of the distribution of environmental state and agent action to avoid single sample noise. Indicates an action, Indicates state, This represents the gradient of the commenter network towards the action. This represents the gradient of the policy network with respect to its own parameters. This represents the network parameters for two evaluators, while This represents the parameters of the actor policy network.

[0137] like Figures 2 to 4 As shown, the simulation results of the proposed method for optimizing the dynamic channel transmission characteristics of a boundary layer meteorological observation aircraft are presented. Figure 2 As shown, the impact of different numbers of scattering clusters and optimization algorithms on average energy efficiency in a DRL-assisted boundary layer air-to-ground wireless communication scenario is illustrated. Simulation results indicate that the proposed algorithm converges rapidly in the early stages of training, followed by a significant improvement in energy efficiency, with minimal fluctuations after stabilization, while traditional algorithms converge slowly and exhibit severe fluctuations. Furthermore, a higher number of scattering clusters indicates a more complex channel, and the proposed algorithm demonstrates a higher energy efficiency, showing a significant advantage. Therefore, it can be concluded that the DRL-assisted boundary layer air-to-ground wireless communication system has stronger self-learning and adaptation capabilities to dynamic channels compared to traditional communication systems, achieving higher energy efficiency.

[0138] like Figure 3 As shown, in the DRL-assisted boundary layer air-to-ground wireless communication scenario, the reward value changes of the proposed algorithm and other algorithms during the training process were compared. It can be seen that the proposed algorithm has a better overall convergence effect, and its reward value is closer to the high-yield range than other algorithms, reflecting the stability of the strategy and the convergence advantage. This indicates that the optimal strategy it converges to is better and can maintain higher returns in long-term training. Therefore, the proposed algorithm is more suitable for this task scenario and can learn high-yield strategies more efficiently to dynamically optimize channel transmission.

[0139] like Figure 4 As shown, this illustrates the impact of different Rice factors on the spatial cross-correlation characteristics of the channel in a DRL-assisted boundary layer air-to-ground wireless communication scenario. Simulation results indicate that when the obstacle distribution between the transmitter and receiver is very sparse and there are few scatterers, i.e., the Rice factor is significantly higher... When the value is very large, the spatial cross-correlation characteristics of the transmission path are significantly greater than those of a channel with very dense scatterers (i.e., (When the value is small). When only non-line-of-sight paths dominated by scatterers exist in the channel, the spatial correlation of the channel decreases significantly, approaching zero. Therefore, it can be concluded that when designing a DRL-assisted boundary layer air-to-ground wireless communication system, the impact of the Rice factor on the time-domain non-stationary characteristics of the channel should be considered.

[0140] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0141] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for optimizing the dynamic channel transmission characteristics of a boundary layer meteorological observation vehicle, characterized in that: The non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains are analyzed in a pre-constructed DRL-based boundary layer meteorological observation aircraft channel model. In the pre-constructed boundary layer meteorological observation aircraft channel model based on DRL, the non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains are combined with the flight energy consumption of the meteorological observation aircraft and the achievable rate in the channel. The energy efficiency optimization objective and constraints of the meteorological observation aircraft in the channel are constructed, and the reward function of the meteorological observation aircraft is defined. Under constraints, the energy efficiency optimization objective is solved, and the solution is used to plan the optimal flight path based on the reward function of the meteorological observation aircraft to obtain the optimal action parameters of the meteorological observation aircraft adapted to non-stationary channels. The method for constructing the boundary layer meteorological observation aircraft channel model based on DRL includes: considering that the scattering cluster is mobile, when the signal emitted by the transmitter reaches the receiver through the reflection path of the scattering cluster, constructing functional expressions for the transmission distance between the transmitter and receiver and the scattering cluster; and constructing a channel complex impulse response functional expression for the transmission path of the signal reflected by the scattering cluster to reach the receiver. When the signal emitted by the transmitter reaches the receiver via a direct path, construct a functional expression for the transmission distance between the transmitter and receiver; construct a channel complex impulse response functional expression for the transmission path from the transmitter to the receiver. The method for constructing a boundary layer meteorological observation vehicle channel model based on DRL is as follows: Establish a three-dimensional space... In a rectangular coordinate system, the line connecting the midpoint of the transmitting antenna array projection and the midpoint of the receiving antenna array is defined as follows: Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The axis is Axis; defined as passing through the midpoint of the transmitter antenna array projection and perpendicular to it. The lines of the plane are Axis; When a signal emitted by a transmitter reaches the receiver after multiple reflections, we model the scatterer as the first reflection after the signal is emitted by the transmitter, and the last reflection before the signal reaches the receiver, respectively by... and It consists of several scatterers, and the complex multipath between the first and last reflections is characterized by a virtual link; the meteorological observation vehicle flies within the boundary layer region, with the horizontal plane constrained by... arrive and arrive Height limit is arrive ; and They are respectively The minimum and maximum coordinates of the axis; and They are respectively The minimum and maximum coordinates of the axis; and They are respectively The minimum and maximum coordinates of the axis; since the DRL algorithm runs in discrete time steps, we divide the total running time into... A discrete segment, wherein the discrete time interval needs to be small enough to satisfy the continuous-time channel state; When the signal emitted by the transmitter reaches the receiver via the reflection path of the scattering cluster, the method for constructing the functional expression of the transmission distance between the transmitter, receiver, and scattering cluster is as follows: Calculate the transmission distance at the transmitter... The root antenna to the first scattering cluster Time-varying transmission distance of each scatterer ; in, , , It is a sufficiently small discrete time interval to accurately capture the time-varying characteristics of the channel. This represents the distance vector from the origin to the first scattering cluster. This represents the distance from the midpoint of the first scattering cluster to the... The offset vector of each scatterer This represents the aircraft coordinates calculated by the DRL algorithm. Indicates the distance from the midpoint of the transmitting antenna to the [missing information]. The distance vector of the root antenna is expressed as: ； The unit distance vector from the spacecraft to the first scattering cluster is represented as: , ； in, and These represent the angles between the line connecting the transmitter and the first scatterer and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; and These represent the azimuth and elevation angles of the transmitting antenna array, respectively. Calculate the receiver's first The root antenna and the last scattering cluster in the first Time-varying transport distance between scatterers ; in, This indicates the distance from the midpoint of the last scattering cluster to the first... The offset vector of each scatterer Indicates the distance from the midpoint of the transmitting antenna to the [missing information]. The distance vector of the root antenna is expressed as: ； The unit distance vector from the last scattering cluster to the receiver is represented as: , ； in, , , and These represent the angles between the line connecting the transmitter and the first scatterer in the last scattering cluster and the horizontal and vertical planes, respectively. It represents the distance between any two lines on the linear antenna at the transmitting end; This indicates the azimuth angle of the transmitting antenna array.

2. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft according to claim 1, characterized in that: The expression for the channel complex impulse response function of the transmission path reaching the receiver after reflection from the scattering cluster is: ； in, Represents Rice factor; To represent a complex number; Represents independent and uniformly distributed random phases. Indicates wavelength. Indicates the carrier frequency. Indicates the rate of light transmission; This represents the propagation delay of the virtual link between two scattering clusters.

3. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft according to claim 2, characterized in that: When the signal emitted by the transmitter reaches the receiver via a direct path, the method for constructing the functional expression of the transmission distance between the transmitter and receiver is as follows: Calculate the transmitter's first The first antenna and receiver Transmission distance between antennas , ; unit distance vector from transmitter to receiver .

4. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft according to claim 3, characterized in that: The expression for the channel complex impulse response function of the transmission path where the signal directly reaches the receiver includes: The expression for the channel complex impulse response function of the transmission path where the signal directly reaches the receiver is: ; in, and These represent the angles between the line connecting the transmitter and receiver and the horizontal and vertical planes, respectively. The expression for the total channel complex impulse response function is: ； in, Indicates path delay. It is an impulse function.

5. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation aircraft according to claim 4, characterized in that: Methods for analyzing the non-stationary characteristics of the transmission path in the spatial, temporal, and frequency domains in a pre-constructed boundary layer meteorological observation aircraft channel model of DRL include: calculating the first... The first antenna and receiver The cross-correlation function in the spatial domain for the transmission paths between the antennas is: ； in, Indicates discrete-time subscripts. The time interval representing the time-varying characteristics of the channel. Represents the expectation function, and This represents the normalized antenna spacing between the transmitter and receiver. Let the above spatial expression in The time-domain autocorrelation function expression is obtained as follows: ; When the signal emitted by the transmitter reaches the receiver after being reflected by the scatterer and directly incident on the receiver, the transmitter's first... The first antenna and receiver The frequency domain correlation function expression for the transmission path between the root antennas is: ; in, Indicates frequency interval.

6. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation vehicles according to claim 5, characterized in that: Considering the flight energy consumption of meteorological observation aircraft and the achievable rate in the channel, the method for constructing the energy efficiency optimization objective and constraints of meteorological observation aircraft in the channel includes: considering the flight energy consumption of meteorological observation aircraft and achievable rate in the channel The expression is: ； ； in, and These represent the constant blade profile power and induced power of the meteorological observation aircraft in hovering state, respectively. This represents the constant power required for ascent or descent; furthermore, Indicates the rotor blade tip velocity. This refers to the average rotor induced speed during hovering; parameter Indicates the fuselage drag ratio. Represents rotor solidity. and These are air density and rotor disk area, respectively; It is the speed of the meteorological observation aircraft; The objective function for optimizing the energy efficiency of meteorological observation aircraft is: ; The conditional constraints corresponding to the state and action spaces are expressed as follows: ； in, This is the initial position of the meteorological observation aircraft. It is a designated area where meteorological observation aircraft can fly. and These are the velocity and position vectors updated by the DRL for the weather vehicle. This is the maximum battery power consumption of a meteorological observation aircraft. This represents the range of acceleration vectors in the body coordinate system of the weather vehicle. The angular velocity of the weather vehicle is used to update the three-dimensional rotation angle.

7. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation vehicles according to claim 6, characterized in that, The reward function is defined as: ； in, It is a reward symbol. It is a penalty for meteorological observation aircraft flying out of the designated area. , and These are three different weight values. It is the maximum achievable rate of the channel.

8. The method for optimizing the dynamic channel transmission characteristics of boundary layer meteorological observation vehicles according to claim 7, characterized in that, Methods for solving energy efficiency optimization objectives under constraints include: using the dual-delay depth deterministic policy gradient algorithm to solve the energy efficiency optimization objectives of meteorological observation aircraft; Initialize the actor and dual evaluator network, and randomly select... The network is updated in mini-batch using samples, and the update method that minimizes the loss is expressed as: ； in, It is a loss function. This represents the parameters of the first evaluator network. This represents the parameters of the second evaluator network. This indicates the total number of samples. This represents the environmental state observed by the agent. Indicates the agent's state The action to choose from, and These are the value estimates of the output state-action sequence in the evaluator network; The expression for softly updating the target network is: ； in, This represents the value of the target commenter network. Indicates that the agent performs an action. The feedback obtained from the environment then guides the agent to learn better actions. This represents the discount factor, which controls the weight of future rewards. This represents the minimum value estimate obtained from the two-evaluator network. Indicates the next environmental state. This represents the target policy for generating the next state. This refers to the exploratory noise added to the actions, allowing the agent to actively explore the environment during training and avoid getting trapped in local optima. This represents the parameters in the two evaluator networks, and the goal is made more stable by delaying updates. The actor network updates its formula according to the delay policy gradient algorithm as follows: ； in, This indicates gradient calculation. This represents the gradient of the network parameters with respect to the actor's policy. Indicates the strategic objective. It is the expectation function, which takes the expectation of the distribution of environmental state and agent action to avoid single sample noise; Indicates an action, Indicates state, This represents the gradient of the commenter network towards the action. This represents the gradient of the policy network with respect to its own parameters. This represents the network parameters for two evaluators, while This represents the parameters of the actor policy network.

Images