A method and system for adaptive flight control of a UAV with variable center of gravity, and a storage medium

Abstract

The application belongs to the field of avionics and flight control technology, and particularly relates to a UAV variable center of gravity adaptive flight control method, system and storage medium; the method comprises the following steps: collecting running state data and inertial measurement data, and monitoring change characteristics in real time; the change characteristics perform dynamic decoupling; based on internal center of gravity offset torque, the net center of gravity coordinate offset is calculated; by adjusting the virtual force arm parameter, the motor control instruction is generated, so that the virtual thrust center and the physical center of gravity coincide; whether there is a liquid sloshing characteristic frequency is detected; the control parameter is dynamically adjusted to compensate for the center of gravity offset and inhibit liquid sloshing; the application provides a UAV variable center of gravity adaptive flight control method, system and storage medium, which can accurately identify the center of gravity online, decouple environmental interference, and actively adapt to the center of gravity change and liquid sloshing.

Description

Technical Field

[0001] This invention belongs to the field of avionics and flight control technology, specifically relating to a method, system and storage medium for adaptive flight control of unmanned aerial vehicles with variable center of gravity. Background Technology

[0002] With the development of the low-altitude economy, multi-rotor drones are increasingly being used in logistics, agricultural and forestry protection, and emergency rescue. In these applications, the payload conditions of drones are often complex and variable. For example, in medical cold chain transportation, blood samples or vaccines may cause a shift in the center of gravity; in pesticide spraying or oil transportation, the swaying of liquid loads can introduce dynamic disturbance torques.

[0003] Existing multi-rotor logistics drones typically employ standard PID control algorithms, which are based on the fundamental assumption that the drone's physical center of gravity (CoG) coincides with its geometric center (Center of Geometry). During flight, if a shift in the center of gravity occurs, the flight control system primarily relies on the integral term (I Term) in the PID controller to eliminate the steady-state error caused by this shift.

[0004] However, this traditional control strategy has significant technical shortcomings when facing complex loads and environmental disturbances: 1. Confusion exists between wind resistance and load identification. Existing algorithms struggle to effectively distinguish between "external continuous wind drag torque" and "internal load offset torque." If correction is solely based on the integral term, when the UAV performs a yaw maneuver to change its course, the direction of the wind drag torque relative to the fuselage coordinate system changes. The cumulative correction value of the integral term cannot respond to this change in a timely manner, often leading to sudden changes in flight attitude or deviation from the intended path.

[0005] 2. Deterioration in dynamic response performance. The accumulation and release of the integral term takes time, which makes the UAV prone to noticeable "angle drop" or "nose-diving" during takeoff or sudden braking. In addition, existing technologies often ignore the impact of load distribution changes on the fuselage's rotational inertia, leading to PID parameter mismatch under full or partial load conditions, which can easily cause oscillations or response lag.

[0006] 3. Motor overload risk. If the center of gravity torque is passively resisted only by differentiating the rotational speed, the motor on the heavily loaded side will be under high load for a long time, greatly reducing the motor's remaining control authority. Once a sudden turbulence occurs or a large maneuver is required, the motor may reach saturation and be unable to provide sufficient torque, causing the drone to lose control and tumble. Summary of the Invention

[0007] Based on the aforementioned problems in the existing technology, the present invention aims to solve these technical problems. The present invention provides a method, system and storage medium for adaptive flight control of unmanned aerial vehicles with variable center of gravity, which can accurately identify the center of gravity online, decouple environmental interference, and actively adapt to changes in the center of gravity and liquid sloshing.

[0008] To achieve the above objectives, the core technical solution adopted by this invention is as follows: a method for adaptive flight control of an unmanned aerial vehicle (UAV) with a variable center of gravity, the method comprising the following steps: The system collects real-time operating status data of each rotor motor of the UAV and inertial measurement data of the fuselage. The operating status data includes rotational speed and current, and the inertial measurement data includes angular velocity and acceleration. When the drone is hovering, control the drone to perform yaw axis rotation and monitor in real time the changes in the thrust distribution of each rotor motor relative to the fuselage coordinate system during the rotation. Based on the aforementioned change characteristics, dynamic decoupling is performed, decomposing the total disturbance torque required to act on the UAV into environmental wind torque and internal center of gravity offset torque; if the orientation of the high-load area in the thrust difference distribution relative to the fuselage coordinate system shifts with the rotation, it is determined to be environmental wind torque; if the orientation of the high-load area relative to the fuselage coordinate system remains unchanged, it is determined to be internal center of gravity offset torque. Based on the determined internal center of gravity offset moment, calculate the net center of gravity coordinate offset after eliminating environmental wind interference; The motor hybrid control matrix of the flight control system is reconstructed online using the net center of gravity coordinate offset. By adjusting the virtual lever arm parameters used by each rotor motor in the hybrid control matrix calculation, an asymmetric motor control command that enables the UAV to self-balance torque is generated, so that the virtual thrust center coincides with the physical center of gravity under zero attitude control command. During flight control, the total mass and moment of inertia of the fuselage are estimated in real time and in parallel based on the total hovering thrust of the UAV, and the spectrum analysis of the inertial measurement data is performed to detect the presence of liquid sloshing characteristic frequencies. Based on the estimated fuselage rotational inertia and the detected characteristic frequency of liquid sloshing, the PID control parameters of the flight controller are dynamically adjusted to compensate for center of gravity shift and suppress liquid sloshing.

[0009] Furthermore, the step of performing dynamic decoupling based on the changing characteristics specifically includes: constructing the fuselage coordinate system of the UAV and recording the thrust values ​​of the four rotor motors in the hovering state in real time; A motor with the maximum thrust or the largest thrust increment is defined as a high-load motor. During the yaw axis rotation of the UAV, the serial number of the high-load motor and its physical position in the fuselage coordinate system are continuously tracked; If the number of the high-load motor changes and its physical orientation in the fuselage coordinate system shifts, so that the vector direction of the thrust difference distribution corresponds to the fixed wind direction in the geographic coordinate system, then it is confirmed as external wind field interference. If the number of the high-load motor remains unchanged and its physical orientation in the fuselage coordinate system is locked and does not change with the yaw angle, then it is confirmed that the center of gravity shift is caused by internal load.

[0010] Furthermore, the step of calculating the net centroid coordinate offset specifically includes: After confirming the internal center of gravity shift, a solution equation based on torque balance is established, in which the sum of the cross product of the thrust of each rotor motor and the lever arm vector of each motor to the geometric center is zero. By combining the real-time thrust values ​​of each rotor motor with the preset airframe geometric parameters, the lateral and longitudinal offsets of the physical center of gravity relative to the geometric center of the fuselage are obtained by solving the aforementioned equations.

[0011] Furthermore, the step of reconstructing the motor hybrid control matrix of the flight control system online using the net center of gravity coordinate offset specifically includes: Define the original mixing matrix, which represents the linear mapping relationship from flight control commands to the output values ​​of each rotor motor, and includes the lever arm parameter item; Based on the net center of gravity coordinate offset, calculate the equivalent virtual lever arm parameters of each rotor motor; For rotor motors close to the physical center of gravity, the corresponding virtual lever arm parameter value is reduced in the torque allocation calculation of the mixing matrix, so that the reference thrust weight allocated to the rotor motor is increased in the reverse calculation of solving the motor output force command. A modified asymmetric hybrid control matrix is ​​generated and injected into the underlying drive module of the flight control system.

[0012] Furthermore, the steps for estimating the total mass of the fuselage and the moment of inertia of the fuselage specifically include: Establish a mapping model between total hover throttle and total lift, and infer the current total mass of the UAV based on the current hover throttle value; Based on the current total mass and the net center of gravity coordinate offset, the moment of inertia of the fuselage on the roll axis and pitch axis is estimated using the parallel axis theorem or a preset mass distribution model.

[0013] Furthermore, the step of dynamically adjusting the PID control parameters of the flight controller specifically includes: Based on the estimated increase in fuselage rotational inertia, the gain of the proportional term in the attitude control loop is increased synchronously by a preset coefficient. When a low-frequency sinusoidal oscillation in the 0.5Hz to 2Hz frequency band is detected in the inertial measurement data, it is determined that there is a liquid sloshing load. When liquid sloshing load is detected, the proportional gain of the angle loop is automatically reduced, while the differential gain of the angular velocity loop is increased, and the control command is low-pass filtered to block positive feedback excitation of liquid sloshing.

[0014] Furthermore, the method also includes a security monitoring step: Real-time monitoring of the output saturation of each rotor motor; If the output of any rotor motor continuously exceeds the preset safety threshold and cannot be eliminated through hybrid control matrix reconstruction, the maximum flight tilt angle and maximum climb speed of the UAV will be forcibly limited to preserve the control margin for attitude stability.

[0015] Furthermore, a variable center of gravity adaptive flight control system for unmanned aerial vehicles (UAVs) is also provided, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to execute the instructions to implement the functions of the following modules: The data acquisition module is used to collect real-time operating status data of each rotor motor of the UAV and inertial measurement data of the fuselage; The decoupling identification module is used to monitor the changes in thrust difference distribution relative to the fuselage coordinate system when the UAV hovers and performs yaw axis rotation, distinguish between environmental wind torque and internal center of gravity offset torque, and calculate the net center of gravity coordinate offset. The hybrid control reconfiguration module is used to calculate the virtual lever arm parameters of each rotor motor online based on the net center of gravity coordinate offset, generate an asymmetric motor hybrid control matrix, and enable the virtual thrust center to adaptively match the physical center of gravity. The parameter adaptive module is used to estimate the fuselage mass and moment of inertia, identify liquid sloshing characteristics, and dynamically adjust the PID parameters of the flight controller accordingly. The flight control execution module is used to drive each rotor motor according to the reconstructed motor hybrid control matrix and the adjusted PID parameters. The UAV variable center of gravity adaptive flight control system is used to execute the above-mentioned UAV variable center of gravity adaptive flight control method.

[0016] Furthermore, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the above-described method. Beneficial effects

[0017] 1. Improved Control Precision and Handling: This invention modifies the center of gravity compensation from the traditional "PID feedback layer" to a "Mixer allocation layer," regenerating the mixing matrix using the online identification results of the physical center of gravity, effectively reducing the static error caused by center of gravity offset. This active compensation mechanism allows pilots to obtain a highly consistent handling feel under no-load, fully loaded, or severely unbalanced load conditions, avoiding the "nodding" phenomenon during takeoff caused by integral term lag.

[0018] 2. Improved wind resistance and bandwidth utilization: Through a "dual-layer decoupling identification" architecture, this invention can effectively isolate external wind interference from internal center-of-gravity torque. This allows the main bandwidth of the PID controller to focus on resisting sudden gusts and high-frequency disturbances, rather than being wasted on maintaining static center-of-gravity balance, significantly improving the flight stability of the UAV under complex weather conditions.

[0019] 3. Effectively prevents liquid sloshing and runaway: For liquid loads in cold chain transportation or plant protection scenarios, this invention introduces a liquid sloshing suppression strategy based on frequency domain analysis. By notch filtering at specific frequencies and dynamic adjustment of PID parameters, the positive feedback loop between liquid sloshing and aircraft attitude is effectively disrupted, reducing the risk of attitude divergence during emergency stops or maneuvers. Attached Figure Description

[0020] To enable those skilled in the art to more clearly and comprehensively understand the technical solutions of the present invention, preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the accompanying drawings are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. In the accompanying drawings: Figure 1 This is an overall flowchart of the UAV variable center of gravity adaptive flight control method provided in the embodiments of the present invention; Figure 2 This is a schematic diagram of the decoupling identification process between environmental wind force moment and internal center of gravity offset moment provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the dynamic reconstruction process of the hybrid control matrix based on the net centroid coordinate offset provided in the embodiment of the present invention; Figure 4 This is a schematic diagram of the functional modules and data flow of the UAV variable center of gravity adaptive flight control system provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the adaptive adjustment process of PID control parameters based on rotational inertia estimation and liquid sloshing identification provided in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in detail with reference to the accompanying drawings: To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, in the description of the present invention, the term "thrust difference distribution" refers to the deviation distribution of the thrust value output by each rotor motor to maintain attitude balance relative to the average thrust; "high-load motor" refers to the motor that outputs the maximum thrust or the largest thrust increment at the current moment; and "virtual lever arm" refers to the distance parameter from the point of action of each motor to the center of gravity preset in the flight control calculation algorithm (Mixer).

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, several preferred embodiments of the invention will be described in detail below with reference to the accompanying drawings. It should be understood that the embodiments described herein are merely for explaining the invention and do not constitute any limitation on its scope of protection. Any modifications, equivalent substitutions, or improvements made based on the spirit and principles of this invention should be included within the scope of protection of this invention.

[0023] Example 1 This embodiment provides a variable center of gravity adaptive flight control method for unmanned aerial vehicles (UAVs), which mainly runs in the UAV's onboard flight control processor. The method adopts a core architecture of "dual-layer decoupling identification + dynamic hybrid control reconfiguration," and the specific process is as follows: Step 1: Data Acquisition and Initial State Establishment The UAV system collects real-time operating status data of each rotor motor and fuselage attitude data.

[0024] Specifically, the real-time rotational speed (RPM) feedback and bus current (I) values ​​of the four motors (denoted as M1 to M4, assumed to be in an X-type layout) are obtained through the electronic speed controller (ESC). At the same time, the three-axis angular velocity, three-axis acceleration, and attitude angle data of the fuselage are obtained through the onboard inertial measurement unit (IMU).

[0025] Before takeoff or during the initial hovering phase, the system performs a rough initial center of gravity estimate. In calm or light wind conditions, the system controls the drone to maintain a horizontal hovering state. At this time, the torque balance equation is established: ∑(Ti × Li) = 0 Where Ti is the thrust of the i-th motor, which is determined by the current value or speed through a pre-calibrated motor model (e.g., Ti = k·ω). 2 The mapping is obtained; Li is the lever arm vector from the i-th motor axis to the geometric center of the UAV. By solving this equation, the current physical center of gravity coordinates (Xcog, Ycog) can be calculated.

[0026] Step 2: Decoupling from environmental interference (core step) To prevent the continuous constant wind force from being misjudged as an internal center of gravity shift, this embodiment introduces dynamic decoupling logic.

[0027] The drone is controlled to perform a slow yaw rotation in place, for example, commanding the drone to rotate 90 degrees to the right, with the rotation rate controlled at 10-30 degrees / second to ensure the stability of data sampling. During the rotation, the flight control system continuously monitors the changes in the thrust distribution of the four motors relative to the fuselage coordinate system.

[0028] The judgment logic is as follows: 1. Wind interference determination (environmental wind torque): If the "high load area" in the thrust difference distribution of the four motors migrates relative to the fuselage coordinate system.

[0029] For example, initially, with the nose facing north (0 degrees), the system detects that the front motors (M1, M2 in the fuselage coordinate system) have high thrust, and the high-load area is at the front of the fuselage (to counteract the northerly wind). When the nose rotates 90 degrees to face east, in order to continue resisting the northerly wind, the left side of the UAV (i.e., the original front) must provide greater thrust. At this time, the system detects that the left-side motors (M1, M4) in the fuselage coordinate system have become high-load motors.

[0030] This means the high-load area has shifted from the "front of the fuselage" to the "left side of the fuselage." The direction of the thrust difference vector is fixed in the geographic coordinate system (pointing towards the wind direction), but rotates in the fuselage coordinate system. In this case, the system determines that the torque difference originates from external environmental disturbances and does not perform center of gravity compensation.

[0031] 2. Determination of center of gravity offset (internal center of gravity offset torque): If the thrust difference distribution of the four motors is "locked" at a specific position in the fuselage coordinate system.

[0032] For example, regardless of the direction the aircraft is pointing, the thrust of the motor (M2) on the left front of the fuselage is always a certain value higher than that of the diagonal motor (M4). This means that the unbalanced torque is bound to the fuselage and does not change with the heading. At this time, the system determines that the center of gravity shift is caused by internal loads.

[0033] Through the above logic, the system can accurately eliminate wind interference and calculate the net center of gravity coordinate offset (ΔX, ΔY) caused only by internal loads.

[0034] Step 3: Dynamic Reconstruction and Mathematical Principles of the Mixer Matrix This is the core innovation of this embodiment. Traditional flight control maps "roll / pitch / yaw / throttle" commands to "motor output" through a fixed mixing matrix M.

[0035] In physics, torque τ = Force × Lever_Arm. Intuitively, a longer lever arm generally corresponds to a larger torque. However, in the flight control distribution algorithm (Mixer), this is a reverse solution process: given the required torque τ (calculated by the PID controller), the required motor thrust F is calculated.

[0036] That is: F = τ / Lever_Arm.

[0037] Therefore, to compensate for the shift of the center of gravity towards a certain motor (such as M2), physically M2 needs to output greater thrust to support the additional weight. At the algorithm level, we need to make the control system "think" that the torque efficiency of M2 has decreased, thereby forcing the controller to allocate more thrust to it.

[0038] The specific implementation steps are as follows: 1. Based on the net center of gravity coordinate offset (ΔX, ΔY) obtained in step 2, calculate the "virtual lever arm parameters" of each motor.

[0039] 2. If the center of gravity moves toward M2, then decrease the value of the virtual lever arm parameter (L_virtual) corresponding to M2 in the mixing matrix.

[0040] 3. According to the formula F_command = τ_desired / L_virtual, when the denominator L_virtual decreases, in order to maintain the same τ_desired, the calculated F_command will increase.

[0041] 4. Generate the corrected asymmetric hybrid control matrix M' and inject it into the flight control layer.

[0042] Through this mathematical inverse adjustment, even if the integral term (I Term) of the PID controller is zero, the motor output will naturally form a torque distribution that counteracts the shift in the center of gravity, thereby releasing the dynamic range of the integral term.

[0043] Step 4: PID gain scheduling based on load characteristics After resolving the static equilibrium issue, the problem of dynamic response consistency also needs to be addressed.

[0044] 1. Inertia estimation and P parameter adjustment: Based on the total throttle input during hovering, the total mass of the drone, Mass_total, is calculated using a pre-stored thrust-weight curve.

[0045] By combining the total mass and center of gravity offset, the roll axis inertia Jxx and pitch axis inertia Jyy of the fuselage are estimated using the parallel axis theorem. If the estimated inertia increases significantly, the system will automatically increase the gain of the proportional terms (P) in the attitude loop and angular velocity loop to offset the increased inertia and maintain the response speed.

[0046] 2. Liquid sloshing suppression and frequency basis: If the drone carries a liquid tank, the sloshing of the liquid will generate a hysteretic interference torque. The system performs real-time spectrum analysis on the IMU angular velocity data.

[0047] For logistics payload containers commonly used by drones (such as 3L-10L square or cylindrical cold chain boxes), according to the swaying modal theory in fluid mechanics, their first-order swaying natural frequencies typically fall within the range of 0.5Hz to 2Hz. This is the engineering reason for choosing this frequency band as the criterion in this invention.

[0048] Action Strategy: When significant oscillating energy is detected in this frequency band, it is determined to be a liquid load. The system automatically executes a "reduce P, increase D" strategy—reducing the P value of the angle loop to decrease sensitivity to sloshing errors, while increasing the D value (differential term) of the angular velocity loop to enhance damping. Furthermore, a notch filter or low-pass filter with a center frequency equal to the sloshing frequency is applied to control commands to prevent the flight control system from generating "positive feedback excitation" to liquid sloshing.

[0049] Step 5: Security Monitoring The system has a safety monitoring layer that continuously monitors the saturation of individual motors. If a motor (e.g., the heavy-load side motor) is consistently above 90% output, it indicates that the control margin in that direction is extremely low. In this case, the system forcibly limits the UAV's maximum flight tilt angle (e.g., to within 15 degrees) and maximum climb rate to ensure that the motor still has residual power for attitude stabilization in the event of sudden gusts of wind.

[0050] Example 2 To better understand the implementation process of this invention, the following description is based on a specific scenario of medical drones transporting blood samples.

[0051] 1. Loading: The nurse placed the blood cooler box, weighing about 3 kg, into the cabin, about 5 cm to the left front.

[0052] 2. Takeoff Self-Check: The drone hovers after taking off. The algorithm detects through current feedback that the current of the left front motor (M2) is significantly higher than that of the right rear motor (M4) on the opposite side.

[0053] 3. Decoupling Verification: The UAV automatically performed a right-hand rotation. The system detected that during the rotation, the current of the M2 motor remained high, with the high-load area located close to the left front of the body coordinate system. The system confirmed that the internal center of gravity was shifted to the left front.

[0054] 4. Parameter Injection: The system automatically updates the Mixer matrix, reducing the virtual lever arm parameter of M2 (e.g., from 0.4m to 0.35m). This results in M2 receiving more thrust under the same pitch / roll commands, automatically balancing the center of gravity.

[0055] 5. Flight Performance: During subsequent flights, the drone encountered crosswinds, but the flight control system only needed minor attitude adjustments to stabilize it. Because the integral term was not occupied by center-of-gravity error and filtering was applied to handle liquid (blood) sloshing, the drone did not veer off course or experience sustained oscillations caused by sloshing. Example 3: This embodiment provides a UAV variable center of gravity adaptive flight control system capable of executing the above-described method. The system's hardware architecture mainly includes an onboard flight control computer (processor and memory), a sensor array, and actuators. Functionally, the system includes: 1. Data Acquisition Module: This module is responsible for communicating with the underlying sensors and drivers. It reads data from the gyroscope and accelerometer at high frequency and obtains telemetry data of the speed and current of each motor from the electronic speed controller (ESC) via UART or CAN bus.

[0056] 2. Decoupling identification module: This is the core computing unit of the system. It contains a state machine that manages the entire process of "hover detection," "rotation command triggering," "data sliding window sampling," and "torque decomposition." This module outputs the physical center of gravity vector after removing wind force by comparing the force characteristics in the fuselage coordinate system and the geographic coordinate system.

[0057] 3. Hybrid Control Reconfiguration Module: This module receives the center of gravity offset from the decoupling identification module and calculates new motor allocation weights in real time. It utilizes the aforementioned reverse lever arm adjustment algorithm to generate an asymmetric hybrid control matrix. This process is completed in milliseconds, ensuring rapid response even when the load shifts.

[0058] 4. Parameter Adaptation Module: This module includes an inertia estimator and a vibration characteristic analyzer. It estimates the mass based on the total thrust and performs an FFT transform on the IMU data to identify the liquid sloshing frequency. Based on a lookup table method, it outputs real-time correction coefficients for the P, I, and D parameters to the PID controller.

[0059] 5. Flight control execution module: This module includes a basic PID controller and the final motor output stage. It receives the desired attitude command from the upper navigation module, calculates the control input using the PID parameters provided by the parameter adaptation module, and distributes the control input to the four motors using the asymmetric matrix provided by the hybrid control reconfiguration module.

[0060] This invention also provides a computer-readable storage medium, such as flash memory, EEPROM, or hard disk, on which a computer program is stored. When the computer program is executed by the onboard processor of the UAV, it can implement all the steps of the UAV variable center of gravity adaptive flight control method described in Embodiment 1 above. Finally, it should be emphasized that the above-described embodiments are only used to illustrate the technical concept and preferred implementation of the present invention, and are not intended to exhaustively describe or limit the scope of protection of the present invention. Any person skilled in the art, after understanding the spirit and core technical solution of the present invention, may make various modifications, equivalent substitutions, or improvements based on the content disclosed in the present invention without departing from the basic principles of the present invention. These obvious modifications or substitutions should all be considered to be included within the scope of protection claimed by the present invention.

Claims

1. A method for adaptive flight control of an unmanned aerial vehicle (UAV) with a variable center of gravity, characterized in that, The method includes the following steps: The system collects real-time operating status data of each rotor motor of the UAV and inertial measurement data of the fuselage. The operating status data includes rotational speed and current, and the inertial measurement data includes angular velocity and acceleration. When the drone is hovering, control the drone to perform yaw axis rotation and monitor in real time the changes in the thrust distribution of each rotor motor relative to the fuselage coordinate system during the rotation. Based on the aforementioned change characteristics, dynamic decoupling is performed, decomposing the total disturbance torque required to act on the UAV into environmental wind torque and internal center of gravity offset torque; if the orientation of the high-load area in the thrust difference distribution relative to the fuselage coordinate system shifts with the rotation, it is determined to be environmental wind torque; if the orientation of the high-load area relative to the fuselage coordinate system remains unchanged, it is determined to be internal center of gravity offset torque. Based on the determined internal center of gravity offset moment, calculate the net center of gravity coordinate offset after eliminating environmental wind interference; The motor hybrid control matrix of the flight control system is reconstructed online using the net center of gravity coordinate offset: An original hybrid control matrix is ​​defined, representing the linear mapping relationship from flight control commands to the output values ​​of each rotor motor, including a lever arm parameter. Based on the net center of gravity coordinate offset, the equivalent virtual lever arm parameter of each rotor motor is calculated. For rotor motors closer to the physical center of gravity, the corresponding virtual lever arm parameter value is reduced in the torque allocation calculation of the hybrid control matrix, increasing the base thrust weight allocated to that rotor motor in the inverse calculation of solving the motor output force command. A corrected asymmetric hybrid control matrix is ​​generated and injected into the underlying drive module of the flight control system to generate asymmetric motor control commands that enable self-balancing of the UAV's torque, so that the virtual thrust center coincides with the physical center of gravity under zero attitude control commands. During flight control, the total mass and moment of inertia of the fuselage are estimated in real time and in parallel based on the total hovering thrust of the UAV, and the spectrum analysis of the inertial measurement data is performed to detect the presence of liquid sloshing characteristic frequencies. Based on the estimated fuselage rotational inertia and the detected characteristic frequency of liquid sloshing, the PID control parameters of the flight controller are dynamically adjusted to compensate for center of gravity shift and suppress liquid sloshing. The method also includes a safety monitoring step: real-time monitoring of the output saturation of each rotor motor; if the output of any rotor motor continuously exceeds a preset safety threshold and cannot be eliminated by reconstructing the hybrid control matrix, the maximum flight tilt angle and maximum climb speed of the UAV are forcibly limited to retain control margin for attitude stability.

2. The adaptive flight control method for unmanned aerial vehicles with variable center of gravity according to claim 1, characterized in that, The step of performing dynamic decoupling based on the changing characteristics specifically includes: Construct the drone's fuselage coordinate system and record the thrust values ​​of the four rotor motors in real time while hovering; A motor with the maximum thrust or the largest thrust increment is defined as a high-load motor. During the yaw axis rotation of the UAV, the serial number of the high-load motor and its physical position in the fuselage coordinate system are continuously tracked; If the number of the high-load motor changes and its physical orientation in the fuselage coordinate system shifts, so that the vector direction of the thrust difference distribution corresponds to the fixed wind direction in the geographic coordinate system, then it is confirmed as external wind field interference. If the number of the high-load motor remains unchanged and its physical orientation in the fuselage coordinate system is locked and does not change with the yaw angle, then it is confirmed that the center of gravity shift is caused by internal load.

3. The adaptive flight control method for unmanned aerial vehicles with variable center of gravity according to claim 1, characterized in that, The steps for calculating the net centroid coordinate offset specifically include: After confirming the internal center of gravity shift, a solution equation based on torque balance is established, in which the sum of the cross product of the thrust of each rotor motor and the lever arm vector of each motor to the geometric center is zero. By combining the real-time thrust values ​​of each rotor motor with the preset airframe geometric parameters, the lateral and longitudinal offsets of the physical center of gravity relative to the geometric center of the fuselage are obtained by solving the aforementioned equations.

4. The adaptive flight control method for a UAV with a variable center of gravity according to claim 1, characterized in that, The steps for estimating the total mass and moment of inertia of the fuselage specifically include: Establish a mapping model between total hover throttle and total lift, and infer the current total mass of the UAV based on the current hover throttle value; Based on the current total mass and the net center of gravity coordinate offset, the moment of inertia of the fuselage on the roll axis and pitch axis is estimated using the parallel axis theorem or a preset mass distribution model.

5. The adaptive flight control method for a UAV with a variable center of gravity according to claim 1, characterized in that, The steps for dynamically adjusting the PID control parameters of the flight controller specifically include: Based on the estimated increase in fuselage rotational inertia, the gain of the proportional term in the attitude control loop is increased synchronously by a preset coefficient. When a low-frequency sinusoidal oscillation in the 0.5Hz to 2Hz frequency band is detected in the inertial measurement data, it is determined that there is a liquid sloshing load. When liquid sloshing load is detected, the proportional gain of the angle loop is automatically reduced, while the differential gain of the angular velocity loop is increased, and the control command is low-pass filtered to block positive feedback excitation of liquid sloshing.

6. A variable center of gravity adaptive flight control system for unmanned aerial vehicles (UAVs), characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to execute the instructions to implement the functions of the following modules: The data acquisition module is used to collect real-time operating status data of each rotor motor of the UAV and inertial measurement data of the fuselage; The decoupling identification module is used to monitor the changes in thrust difference distribution relative to the fuselage coordinate system when the UAV hovers and performs yaw axis rotation, distinguish between environmental wind torque and internal center of gravity offset torque, and calculate the net center of gravity coordinate offset. The hybrid control reconfiguration module is used to calculate the virtual lever arm parameters of each rotor motor online based on the net center of gravity coordinate offset, generate an asymmetric motor hybrid control matrix, and enable the virtual thrust center to adaptively match the physical center of gravity. The parameter adaptive module is used to estimate the fuselage mass and moment of inertia, identify liquid sloshing characteristics, and dynamically adjust the PID parameters of the flight controller accordingly. The flight control execution module is used to drive each rotor motor according to the reconstructed motor hybrid control matrix and the adjusted PID parameters; the UAV variable center of gravity adaptive flight control system is used to execute the UAV variable center of gravity adaptive flight control method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the UAV variable center of gravity adaptive flight control method as described in any one of claims 1 to 5.

Images