An integrated selection, scheduling, and control system for vertical take-off and landing unmanned aerial vehicles (UAVs).
Through the coordinated linkage of multi-dimensional data perception and dynamic scheduling planning, the system compensates for center of gravity shift and wind field disturbance in real time, solving the control lag problem of vertical take-off and landing UAVs under mission load changes and complex wind field environments, and achieving smooth transition and stable flight.
Patent Information
- Application Number
- CN202611490314.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-23
- Publication Date
- 2026-11-17
AI Technical Summary
Existing vertical take-off and landing (VTOL) unmanned aerial vehicle (UAV) systems suffer from control response lag due to center of gravity shift and wind field changes when carrying different mission payloads, leading to stall or loss of control of the aircraft in complex environments.
The system employs a multi-dimensional data sensing module to collect mission requirements and meteorological data in real time. It calculates the center of gravity offset and wind field feedforward data through a mission-environment coupling selection module, generates obstacle avoidance routes in conjunction with a dynamic four-dimensional scheduling and planning module, and performs feedforward compensation in a transitional adaptive control module to dynamically adjust rotor thrust distribution.
It effectively overcomes the response lag problem of traditional control systems, and realizes smooth transition and stable flight of vertical take-off and landing UAVs in complex mission configurations and harsh weather environments, thereby improving flight safety and control robustness.
Smart Images

Figure CN123113775A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control and cluster scheduling technology for unmanned aerial vehicles (UAVs), specifically to an integrated selection, scheduling, and control system for vertical take-off and landing (VTOL) UAVs. Background Technology
[0002] Vertical takeoff and landing (VTOL) drones combine the convenience of VTOL takeoff and landing of multi-rotor drones with the long endurance of fixed-wing drones, making them widely used in complex mission scenarios such as emergency rescue and material delivery. In actual engineering operations, drones often need to carry payloads of different masses and sizes (such as electro-optical pods, first-aid supply boxes, etc.) according to mission requirements, and the airspace where missions are carried out is often accompanied by complex multi-dimensional wind field changes.
[0003] However, existing vertical takeoff and landing (VTOL) unmanned aerial vehicle (UAV) systems suffer from severe data isolation between configuration selection, flight path scheduling, and underlying flight control, leading to the following core technical problems: Existing underlying flight control systems typically employ a fixed-gain feedback control architecture based on standard unloaded state tuning. During the initial hardware configuration phase, when the UAV carries payloads with different physical properties, the actual mass of the aircraft inevitably deviates from its three-dimensional center of gravity coordinates, generating an inherent static offset torque. Simultaneously, existing flight path scheduling systems only perform static obstacle avoidance in three-dimensional space, failing to convert wind shear and turbulent wind speed vectors along the flight path into time-series feedforward compensation data for transmission to the airborne system. Due to the lack of prior data on center of gravity offset and wind field feedforward data, when a VTOL UAV performs the "transition phase" from VTOL mode to fixed-wing level flight mode in localized strong winds, the aircraft faces complex nonlinear aerodynamic changes and dynamic coupling. At this point, the pure error feedback mechanism of the traditional attitude controller has a significant response lag. The control system can only passively increase the thrust of the corresponding side rotor to compensate after the attitude has undergone a substantial deflection. Under the superposition of sudden wind field disturbance and static center of gravity offset torque, this passive compensation is very likely to cause the rotor motor output speed to reach the physical saturation limit rapidly, which in turn causes lift loss, pitch angle and roll angle divergence in the transition phase, and ultimately leads to the UAV stalling and losing altitude or even completely losing control and crashing. Summary of the Invention
[0004] To achieve the above objectives, the present invention provides the following technical solution: This invention provides an integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs), comprising: a multi-dimensional data sensing module for real-time acquisition of mission requirement data, three-dimensional meteorological data of the target airspace, and hardware status data of candidate UAVs in the fleet; The task-environment coupling selection module is communicatively connected to the multi-dimensional data perception module. It is used to input the task requirement data and the three-dimensional meteorological data into a preset coupling degree evaluation model to calculate the matching degree of the candidate UAV, determine the target vertical take-off and landing UAV, and output the payload configuration parameters of the target vertical take-off and landing UAV. The payload configuration parameters include the center of gravity offset. The dynamic four-dimensional scheduling and planning module is communicatively connected to the task-environment coupled selection module. It is used to constrain the flight envelope of the target vertical take-off and landing UAV based on the payload configuration parameters, and generate a time-stamped spatial obstacle avoidance route in combination with the three-dimensional meteorological data. At the same time, it extracts the wind field feedforward data along the spatial obstacle avoidance route. The transitional adaptive control module, deployed on the airborne end of the target VTOL UAV, is communicatively connected to the mission-environment coupling selection module and the dynamic four-dimensional scheduling and planning module. During the transition phase when the target VTOL UAV switches from vertical takeoff mode to fixed-wing level flight mode according to the spatial obstacle avoidance route, the module inputs the center of gravity offset and the wind field feedforward data as feedforward compensation terms to the airborne active disturbance rejection controller to dynamically adjust the thrust distribution of the target VTOL UAV rotor, thereby suppressing attitude disturbances.
[0005] Furthermore, the multi-dimensional data sensing module includes an air-ground collaborative meteorological inversion unit, which includes a ground meteorological interface, an airborne sensor link, and a spatial interpolation processor. The multi-dimensional data sensing module acquires the measured wind direction and speed around the target airspace through the ground meteorological interface, and receives the wind speed vector of the airborne probe of the UAV in real time through the airborne sensing link; the spatial interpolation processor combines the pre-stored digital elevation model of the target airspace and uses a spatial Gaussian Markov random field to reconstruct the three-dimensional grid of each measured sampling point, and calculates and generates the wind shear gradient and turbulence intensity matrix that are discretely distributed along the vertical height, as the three-dimensional meteorological data of the target airspace. The hardware status data is dynamically extracted through the fleet health assessment unit built into the multi-dimensional data perception module; The fleet health assessment unit acquires real-time data on the AC impedance of the power battery cells, the high-order harmonic distortion of the motor phase current, and the strain monitoring data of the arm structure for each candidate UAV via the airborne bus. Based on the battery electrochemical equivalent circuit model and the motor thermal accumulation equation, the battery static state of charge is converted into a dynamic available peak power considering the maximum discharge rate constraint, and the motor harmonic distortion and structural strain are converted into attitude control margin reduction factors. The dynamic available peak power and the attitude control margin reduction factors together constitute a dynamic performance degradation index characterizing the hardware status data of the candidate UAV.
[0006] Furthermore, the task requirement data includes the physical attributes of the task payload and the spatiotemporal constraints of the task operation; The multi-dimensional data perception module converts the task requirement data into the physical boundary envelope of the UAV through a task parser. The specific execution steps include: Based on the mass distribution and geometric dimensions of the mission payload, the spatial eccentricity of the mission payload relative to the UAV's preset reference coordinate system in the mounted state is calculated, and an assembly eccentricity vector is generated. At the same time, the sensitivity threshold of the mission payload to the external environment is analyzed to generate the flight overload limit range and the upper limit of the airframe vibration acceleration. The assembly eccentricity vector, the flight overload limit range, and the upper limit of the airframe vibration acceleration are bound as the boundary constraints of the mission requirement data. The multidimensional data sensing module also includes a time-sensitive network alignment unit; The time-sensitive network alignment unit establishes a sliding observation window based on a unified clock source for the three-dimensional meteorological data, hardware status data, and task requirement data with different update frequencies. Within each sliding observation window, an adaptive filtering algorithm is used to perform timestamp interpolation and extrapolation compensation on asynchronously arriving meteorological inversion data points and airborne sensor telemetry data points. When a sampling link loss is detected, a confidence weight allocation model based on a time decay factor is introduced to output a multidimensional sensing feature tensor with a unified time scale and confidence label.
[0007] Furthermore, the preset coupling degree evaluation model includes a multimodal aerodynamic energy consumption estimation unit; The multimodal aerodynamic energy consumption estimation unit establishes corresponding nonlinear dynamic sub-models for each candidate UAV, including rotor vertical takeoff and landing mode, tilt and tail-seat transition mode, and fixed-wing cruise mode. It extracts wind shear gradient and ambient air density from the three-dimensional meteorological data, combines them with dynamic performance degradation indices from the hardware status data, and performs piecewise integration through each nonlinear dynamic sub-model to calculate the expected energy consumption for vertical climb, wind resistance transition, and cruise when the candidate UAV executes the mission requirements data. The three are then summed to obtain a total flight energy consumption estimate, which is then converted into an energy dissipation penalty term for evaluating the matching degree. The task-environment coupling selection module has a built-in multi-objective decision optimization unit; The multi-objective decision optimization unit performs matrix mapping between the payload quality and timeliness constraints in the task requirement data and the maximum effective payload and cruise speed of each candidate UAV to generate a task performance index; and extracts the mechanism characteristics of the candidate UAVs resisting the maximum turbulence intensity in the three-dimensional meteorological data to generate an aerodynamic stability index; and constructs a three-dimensional Pareto front with the joint optimization objectives of maximizing the task performance index, maximizing the aerodynamic stability index, and minimizing the energy dissipation penalty term. The three-dimensional Pareto front is solved using a non-dominated sorting genetic algorithm. Based on a preset scene preference weight factor, the unique optimal solution is selected from multiple candidate solutions in the Pareto optimal solution set as the target vertical take-off and landing UAV.
[0008] Furthermore, the task-environment coupling selection module also includes a configuration assembly geometry solver; The configuration assembly geometry solver stores the three-dimensional digital model of each candidate UAV and the standard empty center of gravity coordinate system; after determining the target VTOL UAV, it calls the available physical mount point space matrix corresponding to the target VTOL UAV. The centroid position and physical contour of the task payload in the task requirement data are obtained, and the optimal mounting node without physical interference is found by using a spatial pose matching algorithm. Based on the torque balance equation, combined with the empty mass of the target VTOL UAV, the actual mass of the mission payload, and the spatial coordinates of the optimal mounting node, the actual center of gravity coordinates of the assembled UAV are calculated. Finally, the three-dimensional spatial difference vector between the actual center of gravity coordinates of the UAV and the standard empty center of gravity coordinates is output as the center of gravity offset. The task-environment coupling selection module also includes a safety margin fuse unit; In the process of calculating the matching degree, the safety margin circuit breaker unit establishes a dynamic wind resistance envelope boundary; compares the maximum wind resistance level of the candidate UAV with the instantaneous wind speed extreme value in the three-dimensional meteorological data in real time; if it is detected that the matching degree of all candidate UAVs in the fleet is lower than the preset minimum safety threshold, or if it is determined that the attitude control margin of the target VTOL UAV drops to zero or negative value in the current extreme weather environment, a system-level circuit breaker command is triggered, the payload configuration parameters are refused to be output, and a task suspension and severe environment alarm signal is sent to the upper management system.
[0009] Furthermore, the dynamic four-dimensional scheduling and planning module includes a dynamic envelope correction unit; The dynamic envelope correction unit is used to take the factory standard flight envelope of the target VTOL UAV as a reference, and extract the center of gravity offset and total payload mass from the payload configuration parameters; recalculate the lower limit of the stall speed of the target VTOL UAV in fixed-wing mode based on the total payload mass, and calculate the maximum anti-roll pitch angle and maximum roll angle limits in vertical take-off mode based on the center of gravity offset; perform envelope fitting on the updated lower limit of stall speed, maximum anti-roll pitch angle, and maximum roll angle limits to generate the flight envelope constraint customized for the current mission; The dynamic four-dimensional scheduling and planning module includes a spatiotemporal joint optimization unit; The spatiotemporal joint optimization unit discretizes the target spatial domain into spatiotemporal grid nodes containing four dimensions: longitude, latitude, altitude, and time. The turbulent areas or heavy rainfall areas in the three-dimensional meteorological data where the wind force exceeds the safety threshold are mapped as time-varying obstacles that evolve dynamically over time. Using the flight envelope constraint as the dynamic kinematic constraint for the UAV to transfer between adjacent spatiotemporal grid nodes, a heuristic spatiotemporal search algorithm is adopted to search for the node sequence that minimizes the overall flight time and energy consumption cost function, while avoiding static terrain obstacles and time-varying obstacles, thereby generating the spatial obstacle avoidance route with timestamps.
[0010] Furthermore, the dynamic four-dimensional scheduling and planning module includes a feedforward wind field slice extraction unit; The feedforward wind field slicing extraction unit performs spatiotemporal alignment slicing along the time-stamped spatial obstacle avoidance route using a preset look-ahead time window, extracting the three-dimensional wind speed vector around the track point at the corresponding time stamp; further, based on the planned heading angle and predicted flight speed of the target VTOL UAV at the track point, it establishes a body-following coordinate system; through a coordinate transformation matrix, it decomposes and projects the three-dimensional wind speed vector onto the longitudinal, lateral, and vertical directions of the body-following coordinate system, generating a three-axis wind disturbance component sequence arranged in a time series, and sending the three-axis wind disturbance component sequence as the wind field feedforward data to the transitional adaptive control module; The dynamic four-dimensional scheduling and planning module also includes a local elastic replanning unit; The local elastic replanning unit monitors the update difference of the three-dimensional meteorological data in real time as the target vertical take-off and landing UAV flies along the time-stamped spatial obstacle avoidance route; When a sudden local wind field causes the expected wind resistance margin of the original flight path to exceed the flight envelope constraint, the original flight path is locally spatially repelled and deformed using an elastic band deformation algorithm, starting from the current flight position, while keeping the terminal time window unchanged or requesting an extension. This generates a locally updated flight path that avoids the sudden wind field, and the wind field feedforward data is updated and sent simultaneously.
[0011] Furthermore, the airborne active disturbance rejection controller includes a model-aided extended state observer (ESO) and a feedforward nonlinear error feedback unit; The model-assisted extended state observer introduces the center of gravity offset into the UAV nominal dynamics model to construct a benchmark observation model that includes the actual load eccentricity, which is used to separate and estimate the unmodeled internal system disturbance from the total disturbance. The feedforward nonlinear error feedback unit receives the internal system disturbance compensation amount output by the extended state observer and simultaneously receives the wind field feedforward data; it converts the wind field feedforward data into aerodynamic feedforward compensation torque, which is directly superimposed with the internal system disturbance compensation amount to form a composite control law output, thereby performing early reverse compensation before the wind field disturbance substantially affects the attitude of the UAV.
[0012] Furthermore, the transition state adaptive control module includes a variable center of gravity thrust distribution matrix solution unit; The variable center of gravity thrust distribution matrix calculation unit calculates the lever arm space vector of each rotor center of the target VTOL UAV relative to the actual center of gravity of the assembled UAV in real time according to the center of gravity offset; based on the updated lever arm space vector, it reconstructs the control distribution matrix that maps the desired control torque to the thrust of each rotor; in the transition phase, when the center of gravity offset causes the airframe to have an inherent pitching or tilting offset torque, the reference thrust output weight of the corresponding compensation side rotor is increased through the control distribution matrix, so as to achieve static mechanical self-balancing of the pitch angle of the airframe without occupying the dynamic adjustment margin of the attitude controller; The transition state adaptive control module includes a pneumatic-dynamic hybrid control transition unit; The aerodynamic-dynamic hybrid control transition unit is used to acquire the real-time airspeed and rotor transition tilt angle of the target vertical takeoff and landing UAV during the transition phase; calculate the aerodynamic control efficiency that the current fixed-wing aerodynamic control surfaces (including ailerons, horizontal stabilizer, and vertical stabilizer) can generate based on the real-time airspeed, and establish a control weight normalization allocation function that changes nonlinearly with the real-time airspeed and the rotor transition tilt angle; dynamically decouple and allocate the total expected control quantity output by the airborne active disturbance rejection controller to the rotor motor speed control command and the aerodynamic control surface deflection angle command according to the control weight normalization allocation function, so as to achieve a smooth and disturbance-free handover of lift and attitude control during the transition from vertical takeoff mode to fixed-wing level flight mode.
[0013] Furthermore, the transition state adaptive control module also includes an anti-saturation dynamic degradation unit; When calculating the target thrust of each rotor according to the composite control law and the control allocation matrix, the anti-saturation dynamic degradation unit monitors the physical saturation of the thrust of each rotor in real time. If it is predicted that a certain rotor will reach its physical speed limit after superimposing the extreme wind field feedforward data, a dynamic degradation protection strategy is triggered: according to the preset control priority matrix, the position control accuracy and altitude control accuracy in three-dimensional space are sacrificed first, and the limited available thrust margin is forcibly concentrated and allocated to the roll and pitch attitude control channels to ensure that the core attitude of the target VTOL UAV does not diverge under extreme wind shear.
[0014] Beneficial effects This invention, through the coordinated linkage of multi-dimensional data perception, mission environment coupling selection, and dynamic four-dimensional scheduling planning, directly introduces the center of gravity offset caused by mission payload and the wind field feedforward data in the four-dimensional flight path into the underlying transitional adaptive control module, reconstructing the control allocation matrix and feedforward composite control law that include actual eccentricity characteristics. This scheme effectively overcomes the response lag problem of traditional pure error feedback control when facing changes in the center of gravity and sudden wind field disturbances, realizing a smooth handover between aerodynamic and dynamic control during the transition from vertical to horizontal flight of a vertical takeoff and landing UAV. At the same time, combined with the anti-saturation dynamic degradation strategy, it prioritizes the mechanical balance of the core attitude of the airframe when the motor thrust reaches the physical limit, thereby effectively improving the transitional flight safety and overall control robustness of the UAV in complex mission configurations and harsh weather environments. Attached Figure Description
[0015] Figure 1 This is a diagram showing the overall architecture of the control system of the present invention; Figure 2 This is an internal logic block diagram of the multidimensional data sensing module of the present invention; Figure 3 This is an internal logic block diagram of the task-environment coupling selection module of the present invention; Figure 4 This is a block diagram of the internal logic of the dynamic four-dimensional scheduling and planning module of the present invention. Detailed Implementation
[0016] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0017] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but includes other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0018] The present invention will now be described in further detail with reference to the accompanying drawings: Example: like Figures 1 to 4 As shown, an integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) includes: The multi-dimensional data perception module is used to collect task requirement data, three-dimensional meteorological data of the target airspace, and hardware status data of candidate UAVs in the fleet in real time. The task-environment coupling selection module communicates with the multi-dimensional data perception module to input task requirement data and three-dimensional meteorological data into a preset coupling degree evaluation model to calculate the matching degree of candidate UAVs, determine the target vertical take-off and landing UAV, and output the payload configuration parameters of the target vertical take-off and landing UAV, wherein the payload configuration parameters include the center of gravity offset. The dynamic four-dimensional scheduling and planning module communicates with the task-environment coupled selection module. It is used to constrain the flight envelope of the target vertical take-off and landing UAV based on the payload configuration parameters, and generate a time-stamped spatial obstacle avoidance route by combining three-dimensional meteorological data. At the same time, it extracts the wind field feedforward data along the spatial obstacle avoidance route. The transitional adaptive control module, deployed on the airborne end of the target VTOL UAV, communicates with the mission-environment coupling selection module and the dynamic four-dimensional scheduling and planning module, respectively. During the transition phase when the target VTOL UAV changes from vertical take-off mode to fixed-wing level flight mode according to the spatial obstacle avoidance route, the module inputs the center of gravity offset and wind field feedforward data as feedforward compensation terms to the airborne active disturbance rejection controller to dynamically adjust the thrust distribution of the target VTOL UAV rotor, thereby suppressing attitude disturbances.
[0019] Furthermore, the specific operation process of the multi-dimensional data perception module is as follows: The multidimensional data sensing module, serving as the front-end data input of the entire control system, comprises an air-to-ground collaborative meteorological inversion unit, a fleet health assessment unit, a mission parser, and a time-sensitive network alignment unit. Through these four units, the multidimensional data sensing module performs parallel physical acquisition, mechanistic feature calculation, and spatiotemporal synchronous fusion of multi-source heterogeneous data.
[0020] The air-to-ground collaborative meteorological inversion unit performs gridded reconstruction of three-dimensional meteorological data for the target airspace. The ground meteorological interface of the unit reads measured wind direction and speed sequences from meteorological stations surrounding the target airspace via TCP / IP protocol. Simultaneously, the airborne sensor link receives wind speed vectors from the airborne probes of an in-flight UAV performing other missions via Mavlink protocol. The spatial interpolation processor reads the target airspace digital elevation model pre-stored in the system database and establishes a three-dimensional Cartesian coordinate system including longitude, latitude, and altitude. The processor divides the three-dimensional Cartesian coordinate system into three-dimensional grid nodes according to a preset spatial resolution, inputting the read measured ground wind direction and speed, as well as the airborne probe wind speed vectors, as known observation points into a spatial Gaussian Markov random field. The processor calculates the spatial covariance matrix between the known observation points and each three-dimensional grid node, performs iterative inversion, and outputs the predicted wind speed vector for each three-dimensional grid node. The spatial interpolation processor further calculates the spatial partial derivatives of the predicted wind speed vectors on adjacent 3D grid nodes to generate vertically layered discrete wind shear gradients; and calculates the root mean square of wind speed fluctuations at each 3D grid node within a sliding time window to generate a turbulence intensity matrix. The vertically layered discrete wind shear gradients and turbulence intensity matrix together constitute the 3D meteorological data.
[0021] The fleet health assessment unit reads the deep electromechanical operating parameters of each candidate UAV and generates hardware status data. Through the candidate UAV's CAN bus, the unit acquires the cell AC impedance spectrum fed back by the battery management system (BMS), the high-order harmonic distortion of the motor phase current fed back by the electronic speed controller (ESC), and structural strain monitoring data fed back by strain gauges deployed at the root of the UAV's arms. The unit inputs the acquired cell AC impedance spectrum into a built-in electrochemical equivalent circuit model to extract the ohmic and polarization resistance parameters of the cells; combined with the battery's discharge cutoff voltage threshold, it calculates the maximum instantaneous discharge current without triggering battery undervoltage protection, and then determines the dynamic available peak power under the corresponding state. Simultaneously, the fleet health assessment unit inputs the high-order harmonic distortion of the motor phase current into a preset motor thermal accumulation equation to calculate the slope of the motor winding temperature rise caused by harmonics, and maps it to the pulsation amplitude of the motor output torque. Combined with arm structural strain monitoring data, it assesses the nonlinear deformation of the frame under extreme stress. The fleet health assessment unit multiplies the pulsation amplitude of the output torque by the nonlinear deformation, and outputs an attitude control margin reduction factor between 0 and 1. The dynamically available peak power and the attitude control margin reduction factor together construct a dynamic performance degradation index characterizing the physical limits of the candidate UAV.
[0022] The mission parser performs spatial dynamic boundary transformation on the received mission requirement data. It parses the externally input mission payload mass distribution parameters and 3D geometric shape model; sets the geometric center of the candidate UAV as the origin of a preset reference coordinate system, and calculates the centroid coordinates of the mission payload's 3D geometric shape model in the mounted state. The mission parser subtracts the origin coordinates of the preset reference coordinate system from the centroid coordinates, outputting the assembly eccentricity vector. The mission parser extracts externally input mission payload environmental sensitivity threshold parameters (such as the shock resistance level of the optical pod or the maximum centrifugal force limit of liquid agents), substitutes these parameters into a preset dynamic overload mapping matrix, solves for and outputs the flight overload limit range (i.e., the maximum positive and negative G-value range) and the upper limit of airframe vibration acceleration that the load can withstand in 3D space. The assembly eccentricity vector, flight overload limit range, and upper limit of airframe vibration acceleration together constitute the boundary envelope with physical constraint significance in the mission requirement data.
[0023] The Time-Sensitive Network (TSN) alignment unit performs clock synchronization and confidence weighting on 3D meteorological data, hardware status data, and mission requirement data with different update frequencies. The TSN alignment unit establishes a sliding observation window within the system based on a Precision Time Protocol (PTP) clock source. The TSN alignment unit identifies the timestamps of each data stream entering the sliding observation window; for meteorological inversion data points and airborne sensor telemetry data points with phase differences, it initiates a Kalman adaptive filtering algorithm to interpolate and extrapolate the timestamps according to the system's main clock step size, aligning all asynchronous data points to the same sampling time. During the timestamp alignment calculation phase, the TSN alignment unit detects the continuity of each sampling link; if it determines that a data link has experienced packet loss or a delay exceeding the sliding observation window length, the TSN alignment unit activates a confidence weight allocation model based on a time decay factor. The confidence weight allocation model calculates an exponential decay coefficient based on the data loss duration and appends the calculated exponential decay coefficient as a confidence label to the data points generated after extrapolation compensation. Finally, the time-sensitive network alignment unit concatenates the time-aligned 3D meteorological data, hardware status data, task requirement data, and confidence labels into a tensor, outputting a multidimensional sensing feature tensor, and then passes the multidimensional sensing feature tensor to the subsequent modules for coupling computation.
[0024] Furthermore, the specific operation process of the task-environment coupling selection module is as follows: The task-environment coupling selection module receives the multidimensional sensing feature tensor output by the multidimensional data sensing module and starts the internal multimodal aerodynamic energy consumption estimation unit, multi-objective decision optimization unit, configuration assembly geometry solver, and safety margin fuse unit to perform serial-parallel collaborative calculation.
[0025] The multimodal aerodynamic energy consumption estimation unit reads the candidate UAV model identifier from the multidimensional sensing feature tensor. It then retrieves the corresponding nonlinear dynamic sub-models for rotor-based vertical takeoff and landing, tilt and tail-sit transition, and fixed-wing cruise modes from the underlying model library. The unit extracts wind shear gradient, ambient air density, and dynamic performance degradation indices from the multidimensional sensing feature tensor. These indices are then input into the nonlinear dynamic sub-models for rotor-based vertical takeoff and landing, tilt and tail-sit transition, and fixed-wing cruise modes, respectively, and perform piecewise integration of flight drag and motor output in the time domain. After the piecewise integration, the unit outputs the predicted vertical climb energy consumption scalar, the predicted wind resistance transition energy consumption scalar, and the predicted cruise energy consumption scalar for the candidate UAV performing external tasks. The multimodal aerodynamic energy consumption estimation unit performs algebraic summation on the vertical climb estimated energy consumption scalar, the wind resistance conversion estimated energy consumption scalar, and the cruise estimated energy consumption scalar to generate a total flight energy consumption estimate. The multimodal aerodynamic energy consumption estimation unit multiplies the total flight energy consumption estimate by a preset battery internal resistance heat dissipation coefficient, converting it into an energy dissipation penalty term for evaluating the matching degree.
[0026] The multi-objective decision-making optimization unit reads the payload mass parameters and time constraint parameters from the multi-dimensional sensing feature tensor. It establishes a matrix mapping mechanism, performing matrix multiplication and normalization on the payload mass parameters and time constraint parameters with the candidate UAV's maximum effective payload parameters and cruise speed parameters to generate a scalar-form mission effectiveness index. The unit extracts the mechanistic characteristic parameters of the candidate UAV's fuselage skin and airfoil structure to resist maximum turbulence intensity, inputs these parameters into a Lyapunov stability function for convergence determination, and calculates the aerodynamic stability index. The unit sets maximizing the mission effectiveness index, maximizing the aerodynamic stability index, and minimizing the energy dissipation penalty term as joint optimization objectives, constructing a three-dimensional Pareto front containing multiple sets of non-dominated solutions within the three-dimensional decision space. Finally, the unit calls a non-dominated sorting genetic algorithm to perform chromosome crossover, mutation, and non-dominated sorting evolutionary solving operations on each node within the three-dimensional Pareto front. After the non-dominated sorting evolutionary algorithm converges, the multi-objective decision optimization unit extracts the scene preference weight factor matrix pre-stored in the system and performs weighted Euclidean distance calculation between the scene preference weight factor matrix and each non-dominated solution within the three-dimensional Pareto front. The multi-objective decision optimization unit selects the unique optimal solution with the smallest weighted Euclidean distance and outputs the candidate UAV identifier corresponding to the unique optimal solution as the target vertical take-off and landing UAV.
[0027] The configuration assembly geometry solver stores the 3D digital models of the candidate UAV's fuselage and the standard empty-load center of gravity coordinate system matrix in the underlying database. After the multi-objective decision optimization unit outputs the target VTOL UAV, the configuration assembly geometry solver calls the spatial matrix of available physical mount points under the fuselage and wings of the target VTOL UAV. The configuration assembly geometry solver extracts the centroid position data and the 3D point cloud of the physical shape envelope of the mission payload from the mission requirement data. The configuration assembly geometry solver initiates a spatial pose matching algorithm to calculate the spatial intersection volume between the 3D point cloud of the physical shape envelope of the mission payload and the 3D digital model of the target VTOL UAV's fuselage at each physical mount point. The configuration assembly geometry solver selects the physical mount points with a spatial intersection volume of zero as the optimal mount nodes that do not cause physical interference. The configuration assembly geometry solver extracts the empty-load mass parameters of the target VTOL UAV, the actual mass parameters of the mission payload, and the 3D spatial coordinates of the optimal mount nodes in the fuselage coordinate system. The configuration assembly geometry solver constructs a static triaxial moment equilibrium equation. It substitutes the unloaded mass parameter, actual mass parameter, and three-dimensional spatial coordinates into the static triaxial moment equilibrium equation to perform moment accumulation calculations. The result of the moment accumulation calculation is divided by the sum of the unloaded mass parameter and the actual mass parameter to obtain the actual center of gravity coordinates of the assembled aircraft. The configuration assembly geometry solver performs a three-dimensional vector subtraction operation, subtracting the origin coordinates of the standard unloaded center of gravity coordinate system from the actual center of gravity coordinates of the aircraft, generating a three-dimensional spatial difference vector. This three-dimensional spatial difference vector is written as the center of gravity offset and appended to the load configuration parameters of the target vertical takeoff and landing UAV for output.
[0028] The safety margin circuit breaker unit reads the multi-dimensional sensing feature tensor and establishes a dynamic wind resistance envelope boundary. It extracts the instantaneous wind speed extrema from the 3D meteorological data and reads the maximum wind resistance level parameter for each candidate UAV. The unit continuously performs numerical subtraction comparisons between the instantaneous wind speed extrema and the maximum wind resistance level parameter in the time domain. It also extracts the attitude control margin value from the coupling degree evaluation model calculation process. When the unit detects that the matching score of all candidate UAVs in the fleet is lower than the system's built-in minimum safety threshold, or calculates and determines that the attitude control margin value of the target VTOL UAV under the current 3D meteorological data is equal to or less than zero, the unit triggers a hardware-level interrupt. The unit generates a system-level circuit breaker command and blocks the system output port, refusing to send load configuration parameters containing the center of gravity offset to subsequent modules. Meanwhile, the safety margin fuse unit sends a task suspension and severe environmental alarm signal to the upper management system via the communication interface, which includes the current fault code, instantaneous wind speed extreme value, and attitude control margin value.
[0029] Furthermore, the specific operation process of the dynamic four-dimensional scheduling and planning module is as follows: The dynamic four-dimensional scheduling and planning module receives the payload configuration parameters output by the task-environment coupled selection module and the multi-dimensional sensing feature tensor output by the multi-dimensional data sensing module. The dynamic four-dimensional scheduling and planning module activates its internal dynamic envelope correction unit, spatiotemporal joint optimization unit, feedforward wind field slice extraction unit, and local elastic replanning unit to perform route calculation and feedforward data extraction.
[0030] The dynamic envelope correction unit reads the factory-standard flight envelope matrix of the target VTOL UAV as the baseline envelope matrix. The unit analyzes the load configuration parameters and extracts the total payload mass parameters and the three-dimensional coordinates of the center of gravity offset. It inputs the total payload mass parameters into the wing lift balance equations and, under the steady-state level flight assumption that gravity equals lift, solves for the lower limit scalar of the stall speed of the target VTOL UAV in fixed-wing mode. The unit inputs the three-dimensional coordinates of the center of gravity offset into the rigid body dynamics attitude solution model to calculate the maximum anti-rollover pitch angle boundary value and the maximum roll angle limit allowed for maintaining three-axis moment balance in vertical takeoff mode. Finally, the unit inputs the solved lower limit scalar of the stall speed, the maximum anti-rollover pitch angle boundary value, and the maximum roll angle limit into the curved envelope algorithm. Polynomial curve fitting is performed in the three-dimensional coordinate system to generate a continuous physical envelope surface, which is then output as a customized flight envelope constraint.
[0031] The spatiotemporal joint optimization unit discretizes the target airspace, generating a spatiotemporal grid node set containing four dimensions: longitude, latitude, altitude, and timestamp. The unit extracts the wind shear gradient matrix and turbulence intensity matrix from the 3D meteorological data, filtering out turbulent grid nodes and heavy rainfall grid nodes whose scalar values exceed the system's built-in safety threshold. The unit then attaches time evolution labels to the selected turbulent and heavy rainfall grid nodes, mapping them as a time-varying obstacle matrix that dynamically shifts with the time coordinate axis.
[0032] The spatiotemporal joint optimization unit uses flight envelope constraints as the state transition decision equation to verify the kinematic and dynamic feasibility of the target VTOL UAV transitioning between adjacent spatiotemporal grid nodes. The spatiotemporal joint optimization unit initiates a heuristic spatiotemporal search algorithm, which performs node expansion and cost accumulation calculations under the premise of avoiding static terrain obstacles in three-dimensional geographic space and avoiding time-varying obstacle matrices on the fourth-dimensional time axis.
[0033] The spatiotemporal joint optimization unit searches for and outputs a continuous spatiotemporal grid node sequence that minimizes the sum of the overall travel time cost and energy cost function. The spatiotemporal joint optimization unit then connects and smooths the continuous spatiotemporal grid node sequences to generate a spatial obstacle avoidance route with timestamps.
[0034] The feedforward wind field slicing extraction unit establishes a preset look-ahead time calculation window and performs spatiotemporal alignment slicing operations along a spatial obstacle avoidance flight path with timestamps. The unit reads 3D meteorological data consistent with the timestamps of each trackpoint and extracts the 3D absolute wind speed vector matrix within the spatial grid surrounding the trackpoint at the corresponding timestamp. The unit also extracts the planned heading angle parameters and predicted flight speed scalars of the target VTOL UAV at the current trackpoint. Finally, the unit establishes a body-following Cartesian coordinate system based on the predicted centroid of the target VTOL UAV as the origin, using the planned heading angle parameters and predicted flight speed scalars.
[0035] The feedforward wind field slice extraction unit calls the direction cosine matrix derivation formula to generate a coordinate transformation matrix that converts the local geographic coordinate system to the aircraft-following Cartesian coordinate system. The unit then uses the coordinate transformation matrix to perform linear algebraic multiplication on the three-dimensional absolute wind speed vector matrix, decomposing and projecting the absolute wind speed vector onto the longitudinal, lateral, and vertical axes of the aircraft-following Cartesian coordinate system. The unit extracts the projected component data, generating a three-axis wind disturbance component sequence arranged continuously in time series. This three-axis wind disturbance component sequence is then encapsulated as wind field feedforward data and transmitted to the transient adaptive control module via the airborne communication bus.
[0036] The local elastic replanning unit receives the latest 3D meteorological data at a preset polling period while the target VTOL UAV is cruising along a timestamped spatial obstacle avoidance route. It then calculates the update difference matrix between the latest received 3D meteorological data and historical meteorological data. The local elastic replanning unit inputs this update difference matrix into the wind resistance margin comparison model. When the local elastic replanning unit calculates that a sudden local wind field causes the expected wind resistance margin of the original route to exceed the flight envelope constraint boundary, it activates the underlying elastic band deformation algorithm.
[0037] The local elastic replanning unit uses the current three-dimensional coordinates of the flight position as the starting point for spatial optimization. While maintaining the time window constraints at the end of the mission unchanged, it introduces a repulsive potential field function on the spatial nodes of the original flight path. The local elastic replanning unit calculates the artificial repulsive moment generated by the sudden wind field environmental mesh. Based on the artificial repulsive moment, the original flight path nodes undergo geometric bending deformation and physical displacement in three-dimensional space towards a safe airspace with lower potential energy.
[0038] The local elastic replanning unit smooths the node trajectories after physical displacement using a cubic spline interpolation algorithm, generating locally updated flight paths that avoid sudden wind fields. Simultaneously, the local elastic replanning unit wakes up the feedforward wind field slice extraction unit to read the locally updated flight path node information and re-executes slicing, coordinate system establishment, and projection decomposition calculations to generate the latest wind field feedforward data matching the locally updated flight path nodes and then distributes it.
[0039] Furthermore, the specific operation procedure of the transient adaptive control module is as follows: The transient adaptive control module receives wind field feedforward data from the dynamic four-dimensional scheduling and planning module, and load configuration parameters including center of gravity offset from the task-environment coupled selection module. The transient adaptive control module then activates its internal model-aided extended state observer, feedforward nonlinear error feedback unit, variable center of gravity thrust distribution matrix solver, aerodynamic-dynamic hybrid control transient unit, and anti-saturation dynamic degradation unit to perform underlying flight control law calculations and actuator command allocation.
[0040] The model-aided extended state observer reads the load configuration parameters including the center of gravity offset. The model-aided extended state observer substitutes the three-dimensional coordinates of the center of gravity offset into the rigid body Newton-Euler equations of the target VTOL UAV, modifies the moment of inertia tensor matrix, and reconstructs the nominal dynamic model with actual eccentricity characteristics.
[0041] The model-aided extended state observer (MAE) processes the real-time attitude angular velocity and attitude angle signals fed back from the airborne inertial measurement unit using a nominal dynamic model with actual eccentricity characteristics. It calculates the total disturbance moment matrix by establishing a nonlinear tracking differentiator. The MAE then extracts known steady aerodynamic damping moments from the total disturbance moment matrix from a pre-set database, extracting and outputting internal system disturbance compensation quantities characterizing the unmodeled dynamics and internal mechanical friction of the system. The feedforward nonlinear error feedback unit receives the internal system disturbance compensation quantities output by the MAE and simultaneously reads the triaxial wind disturbance component sequence from the wind field feedforward data.
[0042] The feedforward nonlinear error feedback unit calls upon a preset aerodynamic aerodynamic moment coefficient derivative matrix, performs linear algebraic matrix multiplication on the triaxial wind disturbance component sequence and the aerodynamic aerodynamic moment coefficient derivative matrix, and maps it to generate an aerodynamic feedforward compensation torque. The feedforward nonlinear error feedback unit performs a same-direction vector addition on the aerodynamic feedforward compensation torque and the internal system disturbance compensation quantity to generate a feedforward composite disturbance compensation term. The feedforward nonlinear error feedback unit directly introduces the feedforward composite disturbance compensation term into the forward channel of the airborne attitude control inner loop, and performs a superposition and accumulation operation with the reference attitude control law obtained through the nonlinear function fhan, outputting the total desired control torque.
[0043] The variable center of gravity thrust distribution matrix calculation unit reads the three-dimensional coordinates of the center of gravity offset. It then extracts the static lever arm space matrix of each rotor mounting center of the target VTOL UAV in the standard unloaded center of gravity coordinate system. Next, the unit performs a three-dimensional spatial vector subtraction operation, subtracting the three-dimensional coordinates of the center of gravity offset from each item in the static lever arm space matrix, and solves for and outputs the dynamic lever arm space vector of each rotor center relative to the actual center of gravity of the assembled UAV. Based on the dynamic lever arm space vector and combined with the dimensionless thrust and torque coefficients of each rotor, the unit reconstructs the underlying control distribution matrix of the system.
[0044] The reconstructed control allocation matrix establishes a mathematical transformation operator that maps the total desired control torque to the thrust of each rotor. During the transition phase from vertical takeoff mode to fixed-wing level flight mode for the target VTOL UAV, the variable center-of-gravity thrust allocation matrix solution unit reads the longitudinal offset component from the three-dimensional coordinates of the center-of-gravity offset. It then writes the constant term, with the opposite algebraic sign to the longitudinal offset component, into the corresponding diagonal element of the control allocation matrix, increasing the reference thrust output weight of the rotor located on the same side of the center-of-gravity offset. Based on the reconstructed control allocation matrix and the total desired control torque, the variable center-of-gravity thrust allocation matrix solution unit performs inverse matrix multiplication to output a basic thrust command that incorporates static mechanical self-balancing characteristics.
[0045] The aerodynamic-dynamic hybrid control transition unit reads the real-time airspeed scalar from the airborne pitot tube and the rotor transition tilt angle from the airborne angle encoder. The unit then substitutes the real-time airspeed scalar into the aerodynamic pressure calculation formula to calculate the current dynamic pressure value, multiplies it by the aerodynamic reference areas of the aileron, horizontal stabilizer, and vertical stabilizer, and calculates the extreme value of the effective aerodynamic control torque generated by the fixed-wing aerodynamic control surfaces at the current airspeed.
[0046] The aerodynamic-power hybrid control transition unit establishes a two-dimensional interpolation lookup table with real-time airspeed scalar and rotor transition tilt angle values as independent variables, and outputs continuously varying normalized allocation coefficients through the lookup table. The aerodynamic-power hybrid control transition unit uses the normalized allocation coefficients to perform command weight decoupling calculations on the basic thrust command and the total desired control torque. The aerodynamic-power hybrid control transition unit multiplies the total desired control torque by the complementary quantity corresponding to the normalized allocation coefficient, mapping it to a rotor motor speed control pulse width modulation (PWM) command; the aerodynamic-power hybrid control transition unit multiplies the total desired control torque by the normalized allocation coefficient value itself, mapping it to an aerodynamic control surface deflection angle controller command, which is then output to the onboard digital servo.
[0047] During the transition phase, the anti-saturation dynamic degradation unit reads the real-time speed pulse feedback signals of each rotor motor, establishes a speed monitoring time window, and calculates the speed margin percentage between the real-time speed and the motor's physical maximum saturation speed. The anti-saturation dynamic degradation unit extracts the feedforward composite disturbance compensation term and the reconstructed control allocation matrix from the forward channel, performs look-ahead calculations, and predicts the target motor speed value for the next control cycle. When the anti-saturation dynamic degradation unit determines that the target motor speed value of any rotor motor is greater than the motor's physical maximum saturation speed, it activates the underlying fixed control priority matrix.
[0048] The anti-saturation dynamic degradation unit physically disconnects the transmission links between the airborne position control outer loop command and the altitude control outer loop command, forcibly resetting the proportional-integral-derivative (PID) gain coefficient of the position control outer loop and the gain coefficient of the altitude control outer loop to zero. The anti-saturation dynamic degradation unit extracts the remaining available thrust extreme values of all rotor motors and, according to the fixed proportion set by the control priority matrix, allocates all available thrust extreme values to the execution links of the roll attitude control inner loop and the pitch attitude control inner loop, performing closed-loop locking of the three-axis attitude reference angle parameters of the airframe.
[0049] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. An integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs), characterized in that, include: The multi-dimensional data perception module is used to collect task requirement data, three-dimensional meteorological data of the target airspace, and hardware status data of candidate UAVs in the fleet in real time. The task-environment coupling selection module is communicatively connected to the multi-dimensional data perception module. It is used to input the task requirement data and the three-dimensional meteorological data into a preset coupling degree evaluation model to calculate the matching degree of the candidate UAV, determine the target vertical take-off and landing UAV, and output the payload configuration parameters of the target vertical take-off and landing UAV. The payload configuration parameters include the center of gravity offset. The dynamic four-dimensional scheduling and planning module is communicatively connected to the task-environment coupled selection module. It is used to constrain the flight envelope of the target vertical take-off and landing UAV based on the payload configuration parameters, and generate a time-stamped spatial obstacle avoidance route in combination with the three-dimensional meteorological data. At the same time, it extracts the wind field feedforward data along the spatial obstacle avoidance route. The transitional adaptive control module, deployed on the airborne end of the target VTOL UAV, is communicatively connected to the mission-environment coupling selection module and the dynamic four-dimensional scheduling and planning module. During the transition phase when the target VTOL UAV switches from vertical takeoff mode to fixed-wing level flight mode according to the spatial obstacle avoidance route, the module inputs the center of gravity offset and the wind field feedforward data as feedforward compensation terms to the airborne active disturbance rejection controller to dynamically adjust the thrust distribution of the target VTOL UAV rotor, thereby suppressing attitude disturbances.
2. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The multi-dimensional data sensing module includes an air-ground collaborative meteorological inversion unit, which includes a ground meteorological interface, an airborne sensor link, and a spatial interpolation processor. The multi-dimensional data sensing module acquires the measured wind direction and speed around the target airspace through the ground meteorological interface, and receives the wind speed vector of the airborne probe of the UAV in real time through the airborne sensing link; the spatial interpolation processor combines the pre-stored digital elevation model of the target airspace and uses a spatial Gaussian Markov random field to reconstruct the three-dimensional grid of each measured sampling point, and calculates and generates the wind shear gradient and turbulence intensity matrix that are discretely distributed along the vertical height, as the three-dimensional meteorological data of the target airspace. The hardware status data is dynamically extracted through the fleet health assessment unit built into the multi-dimensional data perception module; The fleet health assessment unit acquires real-time data on the AC impedance of the power battery cells, the high-order harmonic distortion of the motor phase current, and the strain monitoring data of the arm structure for each candidate UAV via the airborne bus. Based on the battery electrochemical equivalent circuit model and the motor thermal accumulation equation, the battery static state of charge is converted into a dynamic available peak power considering the maximum discharge rate constraint, and the motor harmonic distortion and structural strain are converted into attitude control margin reduction factors. The dynamic available peak power and the attitude control margin reduction factors together constitute a dynamic performance degradation index characterizing the hardware status data of the candidate UAV.
3. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 2, characterized in that, The task requirement data includes the physical properties of the task payload and the spatiotemporal constraints of the task operation. The multi-dimensional data perception module converts the task requirement data into the physical boundary envelope of the UAV through a task parser. The specific execution steps include: Based on the mass distribution and geometric dimensions of the mission payload, the spatial eccentricity of the mission payload relative to the UAV's preset reference coordinate system in the mounted state is calculated, and an assembly eccentricity vector is generated. At the same time, the sensitivity threshold of the mission payload to the external environment is analyzed to generate the flight overload limit range and the upper limit of the airframe vibration acceleration. The assembly eccentricity vector, the flight overload limit range, and the upper limit of the airframe vibration acceleration are bound as the boundary constraints of the mission requirement data. The multidimensional data sensing module also includes a time-sensitive network alignment unit; The time-sensitive network alignment unit establishes a sliding observation window based on a unified clock source for the three-dimensional meteorological data, hardware status data, and task requirement data with different update frequencies. Within each sliding observation window, an adaptive filtering algorithm is used to perform timestamp interpolation and extrapolation compensation on asynchronously arriving meteorological inversion data points and airborne sensor telemetry data points. When a sampling link loss is detected, a confidence weight allocation model based on a time decay factor is introduced to output a multidimensional sensing feature tensor with a unified time scale and confidence label.
4. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 3, characterized in that, The preset coupling degree evaluation model includes a multimodal aerodynamic energy consumption estimation unit; The multimodal aerodynamic energy consumption estimation unit establishes corresponding nonlinear dynamic sub-models for each candidate UAV, including rotor vertical takeoff and landing mode, tilt and tail-seat transition mode, and fixed-wing cruise mode. It extracts wind shear gradient and ambient air density from the three-dimensional meteorological data, combines them with dynamic performance degradation indices from the hardware status data, and performs piecewise integration through each nonlinear dynamic sub-model to calculate the expected energy consumption for vertical climb, wind resistance transition, and cruise when the candidate UAV executes the mission requirements data. The three are then summed to obtain a total flight energy consumption estimate, which is then converted into an energy dissipation penalty term for evaluating the matching degree. The task-environment coupling selection module has a built-in multi-objective decision optimization unit; The multi-objective decision optimization unit performs matrix mapping between the payload quality and timeliness constraints in the task requirement data and the maximum effective payload and cruise speed of each candidate UAV to generate a task performance index; and extracts the mechanism characteristics of the candidate UAVs resisting the maximum turbulence intensity in the three-dimensional meteorological data to generate an aerodynamic stability index; and constructs a three-dimensional Pareto front with the joint optimization objectives of maximizing the task performance index, maximizing the aerodynamic stability index, and minimizing the energy dissipation penalty term. The three-dimensional Pareto front is solved using a non-dominated sorting genetic algorithm. Based on a preset scene preference weight factor, the unique optimal solution is selected from multiple candidate solutions in the Pareto optimal solution set as the target vertical take-off and landing UAV.
5. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 4, characterized in that, The task-environment coupling selection module also includes a configuration assembly geometry solver; The configuration assembly geometry solver stores the three-dimensional digital model of each candidate UAV and the standard empty center of gravity coordinate system; after determining the target VTOL UAV, it calls the available physical mount point space matrix corresponding to the target VTOL UAV. The centroid position and physical contour of the task payload in the task requirement data are obtained, and the optimal mounting node without physical interference is found by using a spatial pose matching algorithm. Based on the torque balance equation, combined with the empty mass of the target VTOL UAV, the actual mass of the mission payload, and the spatial coordinates of the optimal mounting node, the actual center of gravity coordinates of the assembled UAV are calculated. Finally, the three-dimensional spatial difference vector between the actual center of gravity coordinates of the UAV and the standard empty center of gravity coordinates is output as the center of gravity offset. The task-environment coupling selection module also includes a safety margin fuse unit; In the process of calculating the matching degree, the safety margin circuit breaker unit establishes a dynamic wind resistance envelope boundary; compares the maximum wind resistance level of the candidate UAV with the instantaneous wind speed extreme value in the three-dimensional meteorological data in real time; if it is detected that the matching degree of all candidate UAVs in the fleet is lower than the preset minimum safety threshold, or if it is determined that the attitude control margin of the target VTOL UAV drops to zero or negative value in the current extreme weather environment, a system-level circuit breaker command is triggered, the payload configuration parameters are refused to be output, and a task suspension and severe environment alarm signal is sent to the upper management system.
6. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The dynamic four-dimensional scheduling and planning module includes a dynamic envelope correction unit; The dynamic envelope correction unit is used to take the factory standard flight envelope of the target vertical take-off and landing UAV as a reference and extract the center of gravity offset and total payload mass from the payload configuration parameters. Based on the total payload mass, the stall speed limit of the target VTOL UAV in fixed-wing mode is recalculated, and the maximum anti-roll pitch angle and maximum roll angle limits in vertical take-off mode are calculated based on the center of gravity offset. The updated stall speed limit, maximum anti-roll pitch angle and maximum roll angle limits are then envelope-fitted to generate the flight envelope constraint customized for the current mission. The dynamic four-dimensional scheduling and planning module includes a spatiotemporal joint optimization unit; The spatiotemporal joint optimization unit discretizes the target spatial domain into spatiotemporal grid nodes containing four dimensions: longitude, latitude, altitude, and time. The turbulent areas or heavy rainfall areas in the three-dimensional meteorological data where the wind force exceeds the safety threshold are mapped as time-varying obstacles that evolve dynamically over time. Using the flight envelope constraint as the dynamic kinematic constraint for the UAV to transfer between adjacent spatiotemporal grid nodes, a heuristic spatiotemporal search algorithm is adopted to search for the node sequence that minimizes the overall flight time and energy consumption cost function, while avoiding static terrain obstacles and time-varying obstacles, thereby generating the spatial obstacle avoidance route with timestamps.
7. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 6, characterized in that, The dynamic four-dimensional scheduling and planning module includes a feedforward wind field slice extraction unit. The feedforward wind field slicing extraction unit performs spatiotemporal alignment slicing along the time-stamped spatial obstacle avoidance route using a preset look-ahead time window, extracting the three-dimensional wind speed vector around the track point at the corresponding time stamp; further, based on the planned heading angle and predicted flight speed of the target VTOL UAV at the track point, it establishes a body-following coordinate system; through a coordinate transformation matrix, it decomposes and projects the three-dimensional wind speed vector onto the longitudinal, lateral, and vertical directions of the body-following coordinate system, generating a three-axis wind disturbance component sequence arranged in a time series, and sending the three-axis wind disturbance component sequence as the wind field feedforward data to the transitional adaptive control module; The dynamic four-dimensional scheduling and planning module also includes a local elastic replanning unit; The local elastic replanning unit monitors the update difference of the three-dimensional meteorological data in real time as the target vertical take-off and landing UAV flies along the time-stamped spatial obstacle avoidance route; When a sudden local wind field causes the expected wind resistance margin of the original flight path to exceed the flight envelope constraint, the original flight path is locally spatially repelled and deformed using an elastic band deformation algorithm, starting from the current flight position, while keeping the terminal time window unchanged or requesting an extension. This generates a locally updated flight path that avoids the sudden wind field, and the wind field feedforward data is updated and sent simultaneously.
8. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 7, characterized in that, The airborne active disturbance rejection controller includes a model-aided extended state observer and a feedforward nonlinear error feedback unit; The model-assisted extended state observer introduces the center of gravity offset into the UAV nominal dynamics model to construct a benchmark observation model that includes the actual load eccentricity, which is used to separate and estimate the unmodeled internal system disturbance from the total disturbance. The feedforward nonlinear error feedback unit receives the internal system disturbance compensation amount output by the extended state observer and simultaneously receives the wind field feedforward data; it converts the wind field feedforward data into aerodynamic feedforward compensation torque, which is directly superimposed with the internal system disturbance compensation amount to form a composite control law output, thereby performing early reverse compensation before the wind field disturbance substantially affects the attitude of the UAV.
9. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 8, characterized in that, The transition state adaptive control module includes a variable center of gravity thrust distribution matrix calculation unit; The variable center of gravity thrust distribution matrix calculation unit calculates the lever arm space vector of each rotor center of the target VTOL UAV relative to the actual center of gravity of the assembled UAV in real time according to the center of gravity offset; based on the updated lever arm space vector, it reconstructs the control distribution matrix that maps the desired control torque to the thrust of each rotor; in the transition phase, when the center of gravity offset causes the airframe to have an inherent pitching or tilting offset torque, the reference thrust output weight of the corresponding compensation side rotor is increased through the control distribution matrix, so as to achieve static mechanical self-balancing of the pitch angle of the airframe without occupying the dynamic adjustment margin of the attitude controller; The transition state adaptive control module includes a pneumatic-dynamic hybrid control transition unit; The aerodynamic-dynamic hybrid control transition unit is used to acquire the real-time airspeed and rotor transition tilt angle of the target vertical takeoff and landing UAV during the transition phase; calculate the aerodynamic control efficiency that the current fixed-wing aerodynamic control surfaces can generate based on the real-time airspeed, and establish a control weight normalization allocation function that changes nonlinearly with the real-time airspeed and the rotor transition tilt angle; dynamically decouple and allocate the total expected control quantity output by the airborne active disturbance rejection controller to the rotor motor speed control command and the aerodynamic control surface deflection angle command according to the control weight normalization allocation function, so as to achieve a smooth and disturbance-free handover of lift and attitude control during the transition from vertical takeoff mode to fixed-wing level flight mode.
10. The integrated selection, scheduling, and control system for vertical takeoff and landing unmanned aerial vehicles (UAVs) according to claim 9, characterized in that, The transition state adaptive control module also includes an anti-saturation dynamic degradation unit; When calculating the target thrust of each rotor according to the composite control law and the control allocation matrix, the anti-saturation dynamic degradation unit monitors the physical saturation of the thrust of each rotor in real time. If it is predicted that a certain rotor will reach its physical speed limit after superimposing the extreme wind field feedforward data, a dynamic degradation protection strategy is triggered: according to the preset control priority matrix, the position control accuracy and altitude control accuracy in three-dimensional space are sacrificed first, and the limited available thrust margin is forcibly concentrated and allocated to the roll and pitch attitude control channels to ensure that the core attitude of the target VTOL UAV does not diverge under extreme wind shear.
