Large fixed-wing unmanned aerial vehicle high-precision launching system

Abstract

The invention discloses a high-precision delivery system for a large fixed-wing unmanned aerial vehicle. The high-precision delivery system comprises an unmanned aerial vehicle platform, a delivery device, a navigation and sensing unit and a control unit, the control unit is used for executing the dynamic air-drop window and flight attitude cooperative control method, and comprises the following steps: calculating a dynamic air-drop starting window in real time through a preset air-drop trajectory model based on unmanned aerial vehicle state information and environmental wind field information obtained by the navigation and sensing unit in real time and pre-stored target point geographic information; after entering the air-drop starting window, the control unit synchronously executes the following operations: controlling the unmanned aerial vehicle platform to execute a predetermined attitude maneuvering action so as to stabilize initial conditions during cargo separation; meanwhile, in the attitude maneuver execution process, the optimal putting opportunity is continuously evaluated according to the putting quality factor Q which is calculated in real time. According to the invention, high-precision launching of the unmanned aerial vehicle can be realized.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control, and more specifically to a high-precision delivery system for large fixed-wing UAVs. Background Technology

[0002] With the large-scale application of drone technology in areas such as material delivery, emergency rescue, and battlefield resupply, large fixed-wing drones, with their advantages of long endurance, strong payload capacity, and stable flight speed, have become the core carrier for long-distance, large-volume delivery missions. However, their inherent flight characteristics also bring inherent challenges to the delivery level: compared to small drones or helicopters, large fixed-wing drones fly faster and have greater inertia. In complex environments such as low altitudes and mountainous areas, they are easily affected by vertical wind shear and terrain disturbances, resulting in significant fluctuations in flight attitude. These factors directly affect the initial conditions for cargo separation, creating natural obstacles to accurate delivery.

[0003] To meet the mission requirement of accurately delivering cargo within the target tolerance range, large fixed-wing UAVs need to be equipped with a high-precision delivery system. This system typically includes four core modules: the UAV platform, the delivery device, the navigation sensor unit, and the control unit. However, existing high-precision delivery systems still suffer from problems in practical applications, such as coarse calculation of the airdrop window, poor adaptability to attitude maneuvers, insufficient accuracy in wind field perception, and unresolved issues related to cargo separation impact and residual disturbances. Therefore, a high-precision delivery system for large fixed-wing UAVs is proposed. Summary of the Invention

[0004] The present invention solves the above-mentioned technical problems through the following technical solutions. The present invention includes an unmanned aerial vehicle platform, a delivery device, a navigation and sensing unit, and a control unit. The control unit is used to execute a dynamic airdrop window and flight attitude coordinated control method, including: Based on the real-time acquisition of UAV status information, environmental wind field information, and pre-stored target point geographic information by the navigation and sensing unit, the dynamic airdrop initiation window is calculated in real time through the preset airdrop trajectory model. The airdrop initiation window is a spatiotemporal region that is the set of all flight states that the UAV can adjust its flight attitude and trigger the drop within a continuous time period, while still ensuring that the cargo falls within the target tolerance range. Upon entering the airdrop launch window, the control unit synchronously performs the following operations: controls the drone platform to execute predetermined attitude maneuvers to stabilize the initial conditions during cargo separation; Meanwhile, during the execution of the attitude maneuver, the optimal delivery timing is continuously evaluated based on the real-time calculated delivery quality factor Q, and the delivery device is triggered to deliver the product when the delivery quality factor Q meets the predetermined conditions.

[0005] Furthermore, the airdrop initiation window is determined through the following calculation process: First, based on UAV dynamics and a pre-defined control law, the UAV trajectory is predicted for a period of time in the future starting from the current moment. This trajectory consists of several discrete future state points. Composition, in which For location, For speed, For time, For height; Secondly, for each future state point This is considered a potential drop point, and the airdrop trajectory model is used to predict the cargo landing point. The specific process of predicting cargo landing point using an airdrop trajectory model includes: Step a: Calculate the total descent time of the cargo in the air. Considering the initial vertical velocity and air resistance after the cargo is released, the following simplified model is used for estimation: ; in, The vertical velocity at the moment of cargo release is determined by the UAV's attitude and velocity, where g is the acceleration due to gravity and k is a dimensionless adjustment coefficient related to the cargo's aerodynamic characteristics. For streamlined goods, the k value is closer to 1, while for high-resistance goods, the k value decreases. Corresponding values ​​can be preset according to different goods. Step b: Calculate the net horizontal displacement of the cargo. This displacement is a combination of the displacement of the cargo as it drifts in the wind field and its own inertial displacement. ; Among them, wind field drift displacement This is obtained by integrating the wind field during the descent process: ; Here, This is obtained by interpolating vertical wind field profile data at a height of... The wind speed vector at that location, This represents the real-time height of the cargo during its descent. Inertial displacement The initial velocity of the cargo upon release is generated under windless conditions: ; Next, calculate the total descent time of the cargo in the air. Net displacement of the object in the horizontal direction Estimated cargo landing point Specifically: ; Finally, filter all that meet the criteria. state point Where R is the allowable landing radius, and the spatiotemporal region formed by these state points is the airdrop initiation window. The geographic coordinates of the target landing point for the mission.

[0006] Furthermore, the predetermined attitude maneuver is a pull-up and attitude stabilization maneuver, specifically including: controlling the UAV to pitch upward around the lateral axis, increasing its angle of attack by a predetermined value to generate an additional upward lift component, which is used to partially offset the initial sinking caused by gravity and wind disturbance at the moment of cargo release.

[0007] Furthermore, the calculation process of the delivery quality factor Q integrates instantaneous delivery accuracy and flight attitude stability, specifically as follows: ; in, To estimate the coordinates of the cargo's landing point based on the current status, This is the current pitch angle of the drone. The target pitch angle for the predetermined attitude maneuver. and These are the weighting coefficients. To prevent small constants with a denominator of zero; The control unit continuously calculates the Q value within the airdrop initiation window and triggers the airdrop when it reaches its maximum value.

[0008] Furthermore, the navigation and sensing unit includes a wind field estimation module based on Doppler lidar. This module estimates the wind speed and direction at different altitudes between the UAV and the ground in real time by emitting a laser beam downwards and forward and analyzing the echo signal, thereby acquiring vertical wind field profile data. This vertical wind field profile data is input into the airdrop trajectory model to accurately calculate wind field drift displacement. The integrand in .

[0009] Furthermore, the control unit includes an adaptive controller for optimizing the parameters of attitude maneuvering actions over complex terrain; The adaptive controller takes the ambient wind shear intensity and terrain undulation as input states, and the pitch angle increment as input. The output action is the pose stability time, and its strategy is trained through reinforcement learning based on the delivery success rate to maximize the expected value of the delivery quality factor Q.

[0010] Furthermore, the delivery device employs an active damping release mechanism. Upon receiving a trigger signal, this mechanism first releases the main lock, and then applies a controllable resisting force to the cargo via a damping arm controlled by a linear servo motor. The calculation formula is: ; Where c is the adjustable damping coefficient. and These are the velocity vectors for the cargo and the drone, respectively; this process lasts for a very short time to smooth the velocity separation between the cargo and the drone.

[0011] Furthermore, during the operation of the active damping release mechanism, the control unit records the body angular velocity disturbance caused by the release. After the damping is released, a pair of... A proportional compensation torque is introduced into the attitude control loop to quickly suppress body oscillations caused by the release shock.

[0012] Compared with existing technologies, this invention has the following advantages: This high-precision delivery system for large fixed-wing UAVs, through the coordinated work of the UAV platform, delivery device, navigation and sensing unit, and control unit, allows the control unit to calculate the dynamic airdrop initiation window based on real-time acquired UAV status, environmental wind field, and target point geographic information, combined with an airdrop trajectory model. This ensures that even when the UAV adjusts its flight attitude to trigger delivery within a continuous time period, the cargo can still fall within the target tolerance range. After entering the airdrop initiation window, the control unit synchronously controls the UAV to perform pull-up and attitude stabilization maneuvers, which can stabilize the initial conditions for cargo separation and partially offset the initial sinking caused by gravity and wind disturbance at the moment of cargo release. Simultaneously, the optimal delivery timing is evaluated and selected by real-time calculation of the delivery quality factor Q, which integrates instantaneous delivery accuracy and flight attitude stability. To improve delivery accuracy, the Doppler lidar wind field estimation module in the navigation and sensing unit can acquire vertical wind field profile data, providing more accurate wind impact prediction for the airdrop trajectory model and further improving the accuracy of landing point prediction. The adaptive controller of the control unit can optimize attitude maneuver parameters based on the environmental wind shear intensity and terrain undulation, and its strategy, trained through reinforcement learning, can maximize the expected value of the delivery quality factor Q, adapting to complex terrain environments. The active damping release mechanism of the delivery device smooths the speed separation process between the cargo and the UAV through controllable damping force, reducing separation impact. At the same time, the control unit introduces a compensating torque after release to quickly suppress airframe oscillations and ensure UAV flight stability. Overall, this significantly improves delivery accuracy and the system's ability to adapt to complex environments, optimizing delivery effects and UAV flight safety. Attached Figure Description

[0013] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0014] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0015] like Figure 1 As shown, this embodiment provides a technical solution: a high-precision delivery system for large fixed-wing UAVs, including a UAV platform, a delivery device, a navigation and sensing unit, and a control unit; The control unit is used to execute a dynamic airdrop window and flight attitude coordinated control method, including: Based on the real-time acquisition of UAV status information, environmental wind field information, and pre-stored target point geographic information by the navigation and sensing unit, the dynamic airdrop initiation window is calculated in real time through the preset airdrop trajectory model. The airdrop initiation window is a spatiotemporal region that is the set of all flight states that the UAV can adjust its flight attitude and trigger the drop within a continuous time period, while still ensuring that the cargo falls within the target tolerance range. Upon entering the airdrop launch window, the control unit synchronously performs the following operations: controls the drone platform to execute predetermined attitude maneuvers to stabilize the initial conditions during cargo separation; Meanwhile, during the execution of the attitude maneuver, the optimal delivery timing is continuously evaluated based on the real-time calculated delivery quality factor Q, and the delivery device is triggered to deliver the product when the delivery quality factor Q meets the predetermined conditions.

[0016] The airdrop initiation window is determined through the following calculation process: First, based on UAV dynamics and a pre-defined control law, the UAV trajectory is predicted for a period of time in the future starting from the current moment. This trajectory consists of several discrete future state points. Composition, in which For location, For speed, For time, For height; Secondly, for each future state point This is considered a potential drop point, and the airdrop trajectory model is used to predict the cargo landing point. The specific process of predicting cargo landing point using an airdrop trajectory model includes: Step a: Calculate the total descent time of the cargo in the air. Considering the initial vertical velocity and air resistance after the cargo is released, the following simplified model is used for estimation: ; in, The vertical velocity at the moment of cargo release is determined by the UAV's attitude and velocity, where g is the acceleration due to gravity and k is a dimensionless adjustment coefficient related to the cargo's aerodynamic characteristics. For streamlined goods, the k-value is closer to 1, while for high-resistance goods, the k-value decreases. Step b: Calculate the net horizontal displacement of the cargo. This displacement is a combination of the displacement of the cargo as it drifts in the wind field and its own inertial displacement. ; Among them, wind field drift displacement This is obtained by integrating the wind field during the descent process: ; Here, This is obtained by interpolating vertical wind field profile data at a height of... The wind speed vector at that location, This represents the real-time height of the cargo during its descent. Inertial displacement The initial velocity of the cargo upon release is generated under windless conditions: ; Next, calculate the total descent time of the cargo in the air. Net displacement of the object in the horizontal direction Estimated cargo landing point Specifically: ; Finally, filter all that meet the criteria. state point Where R is the allowable landing radius, and the spatiotemporal region formed by these state points is the airdrop initiation window. The geographic coordinates of the target landing point for the mission; First, based on UAV dynamics and preset control laws, the UAV trajectory composed of discrete state points is predicted over a period of time. Then, each state point is regarded as a potential drop point. The landing point is calculated in three steps: cargo descent time, composite horizontal net displacement, and estimated cargo landing point, using an airdrop trajectory model. Finally, all state points with a landing point distance ≤ the allowable landing radius R are selected to form a dynamic airdrop initiation window.

[0017] Traditional airdrop systems often have fixed drop points pre-set. However, in actual flight, the speed and altitude of drones are easily affected by airflow disturbances, and the environmental wind field changes in real time, making it easy for fixed-point drops to deviate from the target. This design predicts discrete state points over a period of time and designs the airdrop initiation window as a dynamic spatiotemporal region, rather than relying on a single current state. This allows for real-time adaptation to changes in drone attitude, wind field disturbances, and other factors, ensuring that the drop decision always aligns with the actual flight scenario.

[0018] Traditional landing point estimation often neglects key factors such as air resistance and stratified wind fields, leading to significant errors. Layered modeling addresses these core influencing factors: when calculating descent time, the vertical velocity at the moment of cargo release is incorporated. (Determined by the drone's attitude) and air resistance (adapted to different cargo aerodynamic characteristics through a dimensionless coefficient k); when calculating the horizontal net displacement, the inertial displacement of the cargo with the drone and the drift displacement caused by the stratified wind field are considered simultaneously; finally, the landing point is directly correlated with the state point position and net displacement. Each step of the calculation is based on physical principles, which significantly reduces the prediction error.

[0019] For example, if a large fixed-wing drone is performing a material delivery mission, the geographic information of the target point is pre-stored as follows: (Cartesian coordinate system), permissible landing radius A future state of a drone. The parameters are: position Horizontal speed (Corresponding to the x and y directions respectively), height Vertical velocity at the moment of cargo release (Negative values ​​are indicated by downward movement); The goods being transported are streamlined materials, with a dimensionless adjustment coefficient k=0.9; gravitational acceleration The vertical wind field profile is obtained through the navigation and sensing unit, and the wind speed vectors at different heights are: Wind speed at an altitude of 500m Wind speed at an altitude of 300m Wind speed at an altitude of 100m Wind speed at 0m altitude .

[0020] Calculate the total descent time of the cargo in the air. : ; Substituting the parameters into the calculation: the molecule is First, calculate the value inside the square root: The square root result is approximately 94, therefore the numerator is... ; The denominator is ;final .

[0021] Calculate the net horizontal displacement of the cargo. : According to the formula The specific calculations are as follows: Inertial displacement The initial velocity of the cargo upon release, under windless conditions, is determined by the following formula: ; Substituting the parameters, the displacement in the x-direction is The displacement in the y direction is ,Right now .

[0022] Wind field drift displacement It is obtained by integrating the wind field during the descent. Since the wind field changes with altitude, the average wind speed is used to simplify the calculation. The cargo falls from 500m to 0m, with an average height of 250m. The corresponding wind speed is obtained through linear interpolation. The formula is ; Substituting the parameters, the displacement in the x-direction is The displacement in the y direction is ,Right now .

[0023] Net displacement synthesis: .

[0024] Estimated cargo landing point And filter: The landing point formula is: ; Substituting the parameters yields .

[0025] Verify if the tolerance requirement is met: Calculate the Euclidean distance between the landing point and the target point, using the formula: .

[0026] Substituting the numerical values, the square root is divided into... The square root result is approximately 27.57m, because That is, it is less than the allowable landing radius R, therefore this state point It has been included in the airdrop launch window.

[0027] Dynamic trajectory prediction adapts to scenarios where wind fields change with altitude; multi-factor modeling (such as considering layered wind fields and cargo aerodynamic characteristics) controls the landing point prediction error to 27.57m, which is much smaller than the allowable radius; quantified tolerance standards clarify the basis for state point selection; and the dimensionless coefficient k adapts to the aerodynamic characteristics of streamlined cargo, fully demonstrating its advantages in accuracy, adaptability, and reliability.

[0028] By adopting a technical approach that goes from dynamic trajectory prediction to multi-physical quantity modeling and then to quantitative tolerance screening, the problems of static decision-making, low accuracy and poor adaptability of traditional airdrop systems are solved. The airdrop initiation window is upgraded from a static point to a dynamically controllable spatiotemporal region, which is the core technical support for realizing high-precision delivery of large fixed-wing UAVs.

[0029] The predetermined attitude maneuver is a pull-up and attitude stabilization maneuver, specifically including: controlling the UAV to pitch upward around the lateral axis, increasing its angle of attack by a predetermined value to generate an additional upward lift component, which is used to partially offset the initial sinking caused by gravity and wind disturbance at the moment of cargo release; The moment the cargo is released from the drone, it will initially sink due to its own weight (vertically downward) and environmental wind disturbances (such as downward airflow), resulting in an initial vertical velocity of the cargo. Excessively large (the value becomes even more negative when downward). The pull-up action generates additional upward lift by increasing the angle of attack, which can directly offset part of the downward combined force of gravity and wind disturbance, making the vertical velocity of the cargo when it leaves the container closer to the ideal value, and avoiding errors caused by the initial sinking being too fast, resulting in the landing point being ahead of or deviating from the vertical direction.

[0030] In traditional airdrops, the drone's attitude is easily affected by airflow fluctuations (such as slight pitch and roll), resulting in significant differences in the initial attitude (such as the angle relative to the drone) of cargo separation between different drops, leading to high dispersion in the final landing point. Attitude stabilization is a predetermined standardized operation that adjusts the drone's attitude to a uniform drop attitude benchmark regardless of minor environmental disturbances, ensuring consistent initial attitude of cargo separation for each drop and significantly reducing accuracy fluctuations between different drop missions.

[0031] In complex wind environments such as low-altitude and mountainous areas, downward gusts (wind disturbances) can exacerbate the initial descent of cargo. Traditional delivery methods without attitude maneuvering are prone to significant deviations in landing point due to wind disturbances. Additional lift is equivalent to adding wind-resistant buffers to cargo separation. Even in the event of sudden downward wind disturbances, additional lift can partially offset the downward force of the wind disturbances, preventing a sudden increase in initial descent velocity and ensuring that delivery accuracy remains stable in complex wind fields.

[0032] For example, a large fixed-wing drone is performing material delivery, targeting a specific point. Permissible landing radius The initial state of the drone when it enters the airdrop launch window is: Horizontal speed ,high Initial angle of attack Initial pitch angle ; Cargo parameters: weight Streamlined shape, k=0.9; Environmental wind disturbance: Sudden downdrafts cause additional downward wind disturbance forces when goods are detached. ; Attitude maneuver parameters: Increment of predetermined angle of attack Target angle of attack Target pitch angle Postural stability time ; Auxiliary aerodynamic parameters: air density Reference area of ​​UAV wings Lift slope The increase in lift coefficient for every 1 radian increase in angle of attack.

[0033] Calculate the additional lift generated by the pull-up action : Lift coefficient It is directly proportional to the angle of attack; as the angle of attack increases... (Needs to be converted to radians:) At that time, the increase in the lift coefficient is: ; Substituting the parameters, we get: .

[0034] The formula for calculating additional lift (positive upward) is: ; Among them, the flight speed of the drone Substituting the parameters, we get: ; Compare the initial sinking of the cargo with and without attitude maneuvers: At the moment of cargo separation, the downward net force consists of the component of gravity and the wind disturbance force, while the upward force is the supporting force of the drone on the cargo (before separation). The additional lift is transferred to the cargo through the drone's attitude, offsetting part of the downward net force. Without attitude maneuvering: Total downward force Substituting, we get: ; Vertical acceleration before cargo separation (The minus sign indicates downward), after Initial vertical velocity after separation preparation time (High sinking speed).

[0035] During attitude maneuvers: Additional lift The combined upward and downward forces Substituting, we get: ; Vertical acceleration Similarly, Afterwards, the initial vertical velocity (The sinking speed is significantly reduced).

[0036] Compare landing point errors: According to the descent time formula Calculate the results for both cases. Landing point: No attitude maneuver: Substituting, we get: ; Combined with horizontal net displacement calculation Landing point However, due to the high sinking speed, the actual vertical offset caused the horizontal landing point to be ahead of the target point, ultimately resulting in a distance of approximately [missing information]. (Still) (Inner, but with low accuracy).

[0037] attitude maneuvering Substituting, we get: ; Horizontal net displacement factor The changes were fine-tuned to Landing point The distance from the target point is approximately The accuracy is improved compared to when there is no attitude maneuver.

[0038] The lifting and stabilizing action uses additional lift to reduce the initial vertical velocity of the cargo from... Optimized to This directly offset approximately The downward net force accounts for a portion of the total downward net force. Meanwhile, the standardized angle of attack increment and attitude stabilization time ensured the consistency of attitude adjustment, ultimately improving the landing accuracy by nearly 20%, fully demonstrating its core value of stabilizing initial conditions, resisting wind disturbance, and improving accuracy.

[0039] The calculation process of the delivery quality factor Q integrates instantaneous delivery accuracy and flight attitude stability, specifically as follows: ; in, To estimate the coordinates of the cargo's landing point based on the current status, This is the current pitch angle of the drone. The target pitch angle for the predetermined attitude maneuver. and These are the weighting coefficients. To prevent small constants with a denominator of zero; The control unit continuously calculates the Q value within the airdrop initiation window and triggers the airdrop when it reaches its maximum value. The above process can avoid the risks of single-dimensional decision-making and ensure the reliability of delivery. Traditional airdrops often fall into the decision-making trap of either / or: if the delivery is triggered only by "the estimated landing point is within the tolerance" (a single accuracy indicator), the delivery may deviate when the drone's attitude fluctuates drastically. Excessive amounts of cargo caused the cargo to be affected by the machine's vibrations during separation, resulting in the actual landing point deviating from the estimated value. If deployment is triggered solely by the attitude approaching the target value (a single attitude indicator), it may be deployed when the attitude is stable but the accuracy is poor, and may easily exceed the tolerance due to sudden wind disturbances.

[0040] By forcibly coupling the two dimensions, Q will be maximized only when the accuracy is high (small denominator, large precision term) and the attitude is stable (small deviation, large exponent term), thus fundamentally avoiding task failure caused by misjudgment in a single dimension.

[0041] The airdrop initiation window only solves the problem of being able to drop, but there are still differences in accuracy and attitude at different times within the window; the setting of Q solves the problem of getting the best drop. By continuously calculating the dynamic changes of Q, the superposition time of the accuracy peak and the attitude stability peak is locked, so that the actual landing point deviation is much smaller than the error of random drop within the window.

[0042] The weights are flexibly adjustable to adapt to the needs of various scenarios and tasks. and The design makes the system adaptable to different scenarios: for precision material delivery, it can increase... Prioritize accuracy; even with slight deviations in attitude, as long as the accuracy is high enough, Q can still meet the standard. Emergency delivery for complex terrain: can increase Prioritizing attitude stability to prevent drone loss of control due to attitude fluctuations, while also ensuring basic accuracy. This flexibility requires no hardware modifications; multiple deployment tasks can be covered simply by adjusting parameters, reducing system adaptation costs.

[0043] Continuing with the previous delivery mission: Target point Permissible landing radius ; Target pitch angle of pull-up and stabilization movements ; Set the calculation parameters for Q: (Precision weight) (Attitude weights) (To prevent the denominator from being zero); After the drone enters the airdrop launch window, select three consecutive moments. The key parameters at each time point are as follows: Predicted landing point Current pitch angle ; Predicted landing point Current pitch angle ; Predicted landing point Current pitch angle .

[0044] Calculate the Q value at each time point and select the optimal timing: Calculate the precision term at each time point : The core of the accuracy parameter is the "estimated Euclidean distance between the landing point and the target point." The smaller the distance, the larger the accuracy parameter value. time: First, calculate the Euclidean distance: : Substitute this into the precision term formula: : time: Euclidean distance calculation: ; Substitute the formula for the precision term: ; time: Euclidean distance calculation: ; Substitute the formula for the precision term: ; Calculate the attitude terms at each time step : The core of the attitude term is the deviation between the current pitch angle and the target pitch angle. The smaller the deviation, the closer the exponent term is to 1, and the larger the value of the attitude term. time: Pitch angle deviation: ; Substitute into the attitude term formula: ; time: Pitch angle deviation: ; Substituting the attitude term into the formula: ; time: Pitch angle deviation: ; Substituting the attitude term into the formula: ; Step 3: Calculate the Q value at each time point and determine the optimal deployment time. according to Calculate separately: : ; : ; : ; Comparing the Q values ​​at the three time points, Maximum, therefore the control unit is Trigger the delivery device at the right moment – ​​this is when the deviation between the estimated landing point and the target point is minimized, and the deviation between the drone's attitude and the target's pitch angle is also minimized, making it the optimal delivery time within the airdrop initiation window.

[0045] The navigation and sensing unit includes a wind field estimation module based on Doppler lidar. This module estimates the wind speed and direction at different altitudes between the UAV and the ground in real time by emitting a laser beam downwards and forwards and analyzing the echo signal, thereby acquiring vertical wind field profile data. This vertical wind field profile data is input into the airdrop trajectory model to accurately calculate wind field drift displacement. The integrand in ; Traditional wind field sensing tends to overlook the change in wind speed with altitude. However, this case uses Doppler lidar to obtain wind speed and direction at different altitude levels and generate vertical wind field profile data. This data can accurately match the wind speed corresponding to the real-time height during the cargo's descent, avoiding the calculation deviation of wind field drift displacement caused by the average wind speed across the entire height. The lidar provides real-time active detection and dynamically updates the vertical wind field profile. Even if low-altitude gusts or airflow shear occur during the airdrop, it can promptly correct the wind speed data, ensuring that the wind field drift displacement calculation always closely matches reality. The vertical wind field profile data is directly used for the wind field drift displacement of the airdrop trajectory model. Calculate, reduce The error can directly reduce the deviation in landing point prediction, providing key data support for high-precision delivery.

[0046] For example: Drone status: a future state point ,Location Horizontal speed ,high ; Cargo parameters: Streamlined materials, dimensionless adjustment coefficient The time for the cargo to descend was calculated. ; Target point: ; Wind field data: Vertical wind field profile detected by lidar (updated in real time): high At that time, wind speed vector (The negative x direction represents headwind, and the positive y direction represents crosswind). high At that time, wind speed vector ; high At that time, wind speed vector ; high At that time, wind speed vector ; Cargo drop height change: simplified to linear change t is the falling time.

[0047] Wind field drift displacement calculation based on vertical wind field profile: Wind field drift displacement The integral of the wind speed vector during the cargo's descent is: ; because Linear variation, combined with layered data from the vertical wind field profile, is integrated piecewise according to height intervals. The wind speed in each segment is the average of the values ​​at both ends of the interval, and the time corresponds to the duration of the height change. Section 1: : Corresponding fall time The average wind speed in this section is: ; The drift displacement in this segment: ; Section 2: : Corresponding fall time The average wind speed in this section is: ; The drift displacement in this segment: ; Segment 3: : Corresponding fall time The average wind speed in this section is: ); The drift displacement in this segment:

[0048] Total wind field drift displacement: ; Accuracy verification and point calculation based on precise wind field data: According to the landing point formula, the cargo landing point... This is the sum of the UAV's state point position, inertial displacement, and wind field drift displacement, where the inertial displacement... : ; ; Substitute the data into the calculation: ; Verify the distance between the landing point and the target point: ; This distance is much smaller than the allowable landing radius. Furthermore, due to the accuracy of the vertical wind field data, the landing point deviation is reduced compared to the case where the stratified wind field is ignored, directly demonstrating the improvement in delivery accuracy of this project.

[0049] The control unit includes an adaptive controller for optimizing the parameters of attitude maneuvers over complex terrain. The adaptive controller takes the ambient wind shear intensity and terrain undulation as input states, and the pitch angle increment as input. The output action is the pose stability time, and its strategy is trained through reinforcement learning based on the delivery success rate to maximize the expected value of the delivery quality factor Q. This study uses environmental wind shear intensity and terrain undulation (maximum slope of the target area) as inputs to adjust key parameters of attitude maneuvering in real time. The stronger the wind shear and the steeper the terrain, the more important it is to increase... Enhance wind resistance and lift, extend Stable attitude is achieved, preventing fixed parameters from failing in complex environments; the controller strategy is trained through reinforcement learning driven by deployment success rate, accumulating experience and continuously refining itself across multiple tasks. and The matching rules enable the expected value of the delivery quality factor Q to continuously improve, rather than being limited to the initial setting.

[0050] The scenario and basic parameters remain the same: Drone core status: Initial pitch angle upon entering the airdrop launch window. Horizontal speed ,high ; Target point: Permissible landing radius ; Q calculation parameters: ; Adaptive controller rule: After reinforcement learning training, the simplified relationship between the output parameters and the input (a linear approximation is commonly used in engineering): Pitch angle increment: Unit: °, S is wind shear intensity, unit: °C T represents the topographic relief, in degrees (°). Stabilization time: (Unit: s).

[0051] Adaptive control performance under two typical operating conditions: Operating Condition 1: Small wind shear + gentle terrain (normal environment): Input parameter: wind shear intensity (Wind speed changes gradually with altitude), terrain undulation (The target area is a plain); Adaptive output calculation: ; ; Attitude and Q-value correlation: Target pitch angle ,because Adapted to gentle terrain, actual pitch angle (deviation ); Estimated landing point deviation Substitute into the Q formula: ; Operating Condition 2: Strong wind shear + steep terrain (complex environment) Input parameter: wind shear intensity (Wind speed varies drastically with altitude), terrain undulation (The target area is a mountainous region); Adaptive output calculation: ; ; Attitude and Q-value correlation: Target pitch angle ,because Extended adaptation to steep terrain disturbances, actual pitch angle (deviation ); Estimated landing point deviation Substitute into the Q formula: ; Compared to the disadvantages of fixed parameters (highlighting the advantages): If traditional fixed parameters are used : In operating condition 2, because Insufficient wind resistance and lift, actual pitch angle , and target posture deviation The predicted landing point deviation increased due to the influence of wind shear. At this point, the Q value is: ; The comparison shows that the adaptive controller performs better in operating condition 2 (complex environment). Much higher than the fixed parameters And the attitude deviation remained stable at After multiple tasks and reinforcement learning, the controller can further refine the formula coefficients (e.g., fine-tune the coefficient of S from 0.8 to 0.85), so that... The improvement reflects the advantages of continuous strategy optimization.

[0052] The delivery device employs an active damping release mechanism. Upon receiving a trigger signal, this mechanism first releases the main lock, and then applies a controllable damping force to the cargo via a damping arm controlled by a linear servo motor. The calculation formula is: ; Where c is the adjustable damping coefficient. and These are the velocity vectors of the cargo and the drone, respectively; this process lasts for a very short time to smooth the velocity separation between the cargo and the drone. A controllable damping force is applied by a damping arm controlled by a linear servo motor, forcing the speed difference between the cargo and the drone to decrease slowly rather than abruptly, thus preventing structural damage or functional failure of high-value cargo (such as precision equipment and fragile materials) due to impact. The instantaneous impact of rigid release will bring a reverse torque to the drone, causing pitch / roll oscillations. The damping force disperses the impact energy by slowly offsetting the speed difference, greatly reducing the interference of the release process on the drone's attitude and ensuring subsequent flight stability. The adjustable damping coefficient c can be flexibly adjusted according to the cargo's mass and aerodynamic characteristics. For heavy cargo (such as heavy materials), c can be increased to enhance the damping effect, while for light cargo (such as lightweight equipment), c can be decreased to avoid excessive damping. Different delivery needs can be adapted without changing the release mechanism.

[0053] The scenario and basic parameters remain the same: Drone status: Horizontal speed when entering the deployment phase (x represents the flight direction, y represents the lateral direction), attitude stable; Cargo parameters: weight (Streamlined materials, same as the previous scenario), their speed is consistent with that of the drone before release. ; Active damping parameters: After adaptation, the damping coefficient (Set according to cargo mass), damping duration (Very short time, does not affect the airdrop window); Traditional rigid release compared to: without damping force, the cargo's speed rapidly decreases due to air resistance the instant it detaches. The speed difference is generated instantaneously.

[0054] Calculation of force and velocity changes during active damping release: Damping force Real-time calculation: The damping force is proportional to the speed difference between the cargo and the drone, but in the opposite direction. ; Initial release Due to air resistance, the speed of the cargo decreased. The speed difference at this point is: ; Substitute into the formula to calculate the damping force: ; Smoothing effect of velocity separation process: Comparing the changes in cargo velocity between active damping and conventional rigid release (taking the x-direction as an example): Traditional rigid release: no damping force, speed from Instantly dropped ( (Time), rate of change of velocity: ; (The negative sign indicates deceleration; drastic deceleration produces a large impact.) Active damping release: Under the action of damping force, the velocity decreases slowly. Time drops to , Time drops to Average rate of change of velocity: ; (The deceleration rate is only 1 / 4 of that of the traditional method, and the impact energy is greatly dispersed.) Cargo impact damage risk assessment: Impact acceleration is directly proportional to the impact force on the cargo. : Traditional rigid release: It far exceeds the impact resistance threshold of precision materials (usually <2000N), making it easily damaged; Active damping release: The impact resistance is below the threshold, and the goods are intact.

[0055] Verification of the impact on UAV attitude: In traditional rigid release, the drastic change in the cargo's velocity exerts a counterforce on the drone (Newton's Third Law), causing a momentary fluctuation in the drone's velocity in the x-direction. This causes pitch angle disturbances. ; In active damping release, the damping force acts slowly, and the peak reaction force is only 1 / 4 of that in the traditional method. (Traditional reaction force ≈ 2400N), drone speed fluctuation in the x direction Pitch angle disturbance This is far below the Q-value requirement for attitude stability (a deviation of <0.5° is sufficient), and does not affect the judgment of the optimal deployment time.

[0056] During the operation of the active damping release mechanism, the control unit records the body angular velocity disturbance caused by the release. After the damping is released, a pair of... A proportional compensation torque is introduced into the attitude control loop to quickly suppress body oscillations caused by the release impact; Even after the active damping is released and the impact has been mitigated, residual angular velocity disturbances in the aircraft (such as slight rotations in the pitch / roll direction) will still occur due to the reaction force from cargo separation. Traditional systems rely on the aircraft's own stability to slowly decay these disturbances; however, this mechanism introduces a mechanism that... A proportional compensation torque actively counteracts disturbances, reducing the oscillation decay time to within 0.5 seconds and preventing the disturbance from continuously affecting subsequent flight; residual oscillations can affect the drone's pitch angle. Deviation from target attitude This, in turn, affects the attitude term of the Q value; the compensating torque can quickly [reduce / improve] the [Q value]. Pull back to the target range to ensure that the attitude deviation is always at a minimum when calculating the Q value, so as not to interfere with the judgment of the optimal delivery time; Compensating torque is recorded From calculating the proportional torque to incorporating it into the existing attitude control loop, no new sensors or actuators are needed. Stability can be improved simply by optimizing the control logic, making it compatible with existing systems and cost-effective.

[0057] For example, the drone's status: horizontal speed before deployment. The attitude control loop gain is normal; Prerequisite for releasing disturbance: Perform active damping release Afterwards, pitch angular velocity disturbances still remain. (Clockwise is positive, meaning the body has a slight tendency to tilt its head down). Compensation parameters: After system calibration, the compensation torque and proportionality coefficient (Determined by the rotational inertia of the UAV), the compensating torque only acts until the angular velocity approaches 0; Traditional comparison: Without compensation, the organism relies on its own damping (damping coefficient). Slowly decay the disturbance.

[0058] Compensation torque calculation and vibration damping effect verification: Calculation of compensating torque: Compensating torque With residual angular velocity disturbance Proportional and opposite in direction (to cancel out disturbances), the formula is: ; Substitution : ; (The negative sign indicates that the torque direction is counterclockwise, unlike clockwise.) Conversely, it directly offsets the downward trend. Comparison of angular velocity decay processes (pitch direction): Uncompensated (traditional method): The decay of the body's angular velocity over time follows the laws of a first-order system, as shown in the formula: ; The pitch direction rotational inertia of the drone Substituting into the equation yields the attenuation coefficient. .

[0059] when Decay to 0.05 When the perturbation is negligible, the solution time is: ; (Slow decay, long-lasting residual disturbance) After the compensating torque is introduced, the angular velocity decay formula becomes (the compensating torque is equivalent to increasing the damping): ; Substitution attenuation coefficient .

[0060] Similarly, it decays to Time: ; If further attenuation is required... When there is compensation No compensation required Improved vibration damping efficiency; Verification of the impact on Q value Residual angular velocity disturbances will cause pitch angle Deviation from target value attitude deviation and Points-related: Without compensation, the cumulative attitude deviation reaches [value] within 0–3 seconds. Substitute the attitude term into the Q value: ; With compensation, the attitude deviation within 0 to 3 seconds is only Attitude item: ; Assuming all precision terms are 11.98 (same as the previous example), then: No compensation ; There is compensation .

[0061] It is evident that the compensation mechanism ensures that the Q value remains within the optimal range, preventing a sharp drop in the Q value due to residual oscillations and thus not affecting the decision-making process for deployment timing.

[0062] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0063] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0064] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A high-precision delivery system for large fixed-wing unmanned aerial vehicles (UAVs), characterized in that: This includes the drone platform, delivery device, navigation and sensing unit, and control unit; The control unit is used to execute a dynamic airdrop window and flight attitude coordinated control method, including: Based on the real-time acquisition of UAV status information, environmental wind field information, and pre-stored target point geographic information by the navigation and sensing unit, the dynamic airdrop launch window is calculated in real time through a preset airdrop trajectory model. Upon entering the airdrop launch window, the control unit synchronously performs the following operations: controls the drone platform to execute predetermined attitude maneuvers to stabilize the initial conditions during cargo separation; Meanwhile, during the execution of the attitude maneuver, the optimal delivery timing is continuously evaluated based on the real-time calculated delivery quality factor Q, and the delivery device is triggered to deliver the product when the delivery quality factor Q meets the predetermined conditions.

2. The high-precision delivery system for large fixed-wing UAVs according to claim 1, characterized in that: The airdrop initiation window is determined through the following process: First, based on UAV dynamics and a pre-defined control law, the UAV trajectory is predicted for a period of time in the future starting from the current moment. This trajectory consists of several discrete future state points. Composition, in which For location, For speed, For time, For height; Secondly, for each future state point This was considered a potential drop point, and the cargo landing point was predicted using an airdrop trajectory model. The specific process of predicting cargo landing point using an airdrop trajectory model includes: Step a: Calculate the total descent time of the cargo in the air. ; Step b: Calculate the net horizontal displacement of the cargo. ; Next, calculate the total descent time of the cargo in the air. Net displacement of the object in the horizontal direction Estimated cargo landing point ; Finally, filter all that meet the criteria. state point Where R is the allowable landing radius, the spatiotemporal region formed by these state points is the airdrop launch window.

3. The high-precision delivery system for large fixed-wing UAVs according to claim 1, characterized in that: The predetermined attitude maneuver is a pull-up and attitude stabilization maneuver, specifically including: controlling the UAV to pitch upward around the lateral axis, increasing its angle of attack by a predetermined value to generate an additional upward lift component, which is used to partially offset the initial sinking caused by gravity and wind disturbance at the moment of cargo release.

4. The high-precision delivery system for large fixed-wing UAVs according to claim 3, characterized in that: The specific process for obtaining the delivery quality factor Q is as follows: ； in, To estimate the coordinates of the cargo's landing point based on the current status, This is the current pitch angle of the drone. The geographic coordinates of the target landing point for the mission. The target pitch angle for the predetermined attitude maneuver. and These are the weighting coefficients. To prevent small constants with a denominator of zero; The control unit continuously calculates the Q value within the airdrop initiation window and triggers the drop when it reaches its maximum value.

5. The high-precision delivery system for large fixed-wing UAVs according to claim 2, characterized in that: The navigation and sensing unit includes a wind field estimation module based on Doppler lidar. The wind field estimation module estimates the wind speed and direction at different altitudes between the UAV and the ground in real time by emitting a laser beam downwards and forwards of the UAV and analyzing the echo signal, thereby obtaining vertical wind field profile data. The vertical wind field profile data was input into the air-dropped trajectory model to calculate the wind field drift displacement. The integrand in .

6. The high-precision delivery system for large fixed-wing UAVs according to claim 1, characterized in that: The control unit includes an adaptive controller for optimizing the parameters of attitude maneuvers over complex terrain. The adaptive controller takes the ambient wind shear intensity and terrain undulation as input states, and the pitch angle increment as input. The output action is the pose stability time, and its strategy is trained through reinforcement learning based on the delivery success rate to maximize the expected value of the delivery quality factor Q.

7. The high-precision delivery system for large fixed-wing UAVs according to claim 1, characterized in that: The delivery device employs an active damping release mechanism. Upon receiving a trigger signal, this mechanism first releases the main lock and then applies a controllable resistance force to the cargo through a damping arm controlled by a linear servo motor.

8. The high-precision delivery system for large fixed-wing UAVs according to claim 1, characterized in that: During the operation of the active damping release mechanism, the control unit records the body angular velocity disturbance caused by the release. After the damping is released, a pair of... A proportional compensation torque is introduced into the attitude control loop to quickly suppress body oscillations caused by the release shock.

Images