A center of gravity perception driven aircraft dynamic regulation method and system

Abstract

The application discloses a kind of gravity perception driven aircraft dynamic regulation methods and systems, it is related to aircraft control technical field.The method synchronously obtains three-axis angular velocity and attitude control integral item sequence, extracts steady-state control torque residual by low-pass filter isolation transient disturbance, generates real-time gravity offset vector by reverse mapping;The offset vector is substituted into balance equation to measure each rotor steady-state tension reference value, solve cross-axis cross-coupling interference and be converted into yaw feedforward compensation component;Measure the remaining thrust margin and dynamic response time constant corresponding to tension reference value, execute high-frequency attitude adjustment proportional attenuation to limited rotor with the two as constraint, construct asymmetric dynamic weight penalty matrix;The penalty matrix is carried out weighted generalized inverse operation with basic structure configuration matrix to generate reconstruction control distribution matrix;Finally, yaw feedforward compensation is superimposed to yaw reference instruction, and attitude control instruction is converted into duty cycle signal and issued for execution in combination with reconstruction distribution matrix.

Description

Technical Field

[0001] This invention relates to the field of aircraft control technology, specifically to a method and system for dynamic control of aircraft driven by center of gravity sensing. Background Technology

[0002] In recent years, with the increasing multidimensionality of aircraft operations, carrying dynamic payloads in the air has become a common operating condition. During actual operation, when an aircraft experiences payload separation, internal mass loss, or unexpected physical displacement of the payload, its total onboard mass and spatial distribution change significantly, leading to abrupt or continuous shifts in the aircraft's center of gravity. This shift not only disrupts the initial aerodynamic balance of the aircraft under calibration but also directly reduces the stability margin of attitude control and the maneuverability response space, posing stringent technical requirements for real-time maintenance of flight status and dynamic control of the underlying actuators.

[0003] Most existing aircraft attitude control schemes are based on the static center of mass assumption. When encountering a sudden change in the center of gravity, it is often treated as an external unmodeled disturbance, relying on the passive integral action of the attitude controller for disturbance rejection. This mechanism, when facing continuous asymmetric loads, maintains the underlying thrust distribution logic unchanged, causing the rotor on the offset side to operate under high load for a long time, which easily leads to electromagnetic heating and thrust saturation. At the same time, the existing control distribution system ignores the difference in anti-torsional drag torque generated by each rotor under asymmetric high speeds. When forcibly correcting pitch or roll attitude, the undistributed aerodynamic drag torque will generate severe cross-axis cross-coupling interference, causing the aircraft to exhibit unexpected yaw and spin phenomena. The attitude maintenance capability and anti-coupling interference capability under complex asymmetric load conditions are significantly insufficient. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a gravity sensing-driven dynamic control method and system for aircraft to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a dynamic control method for an aircraft driven by center of gravity perception, comprising the following steps: S1, synchronously acquiring the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller; performing low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances; extracting the steady-state control torque residual characterizing the asymmetric load; combining the three-axis angular velocity sequence, back-mapping the steady-state control torque residual to the body coordinate system to calculate and generate a real-time center of gravity offset vector; S2, substituting the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor; based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, solving the cross-axis cross-coupling caused by the asymmetric thrust distribution. The interference is converted into a reverse yaw feedforward compensation component; S3, calculate the steady-state thrust reference value and the remaining thrust margin of the extreme saturation thrust of each rotor, and map them to generate the dynamic response time constant of the corresponding operating point; use the remaining thrust margin and dynamic response time constant as joint constraints, perform high-frequency attitude adjustment proportional decay on rotors with low margin and high time constant state, and construct an asymmetric dynamic weight penalty matrix; S4, perform weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the basic structure configuration matrix to generate a reconstructed control allocation matrix; superimpose the yaw feedforward compensation component onto the yaw control channel reference command, and jointly reconstruct the control allocation matrix to convert the three-axis attitude control command into an independent duty cycle signal and send it to each rotor actuator.

[0006] Preferably, the specific process of simultaneously acquiring the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller, performing low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances, and extracting the steady-state control torque residual characterizing the asymmetrical load is as follows: Read the integral term output sequences of the pitch and roll channels of the inner loop attitude controller; perform time-domain sliding window analysis on the three-axis angular velocity sequence to extract the high-frequency pulsating boundary frequency characterizing gust disturbances and structural resonance characteristics; set the high-frequency pulsating boundary frequency as the cutoff frequency reference of the adaptive digital filter; import the integral term output sequence into the adaptive digital filter to filter out high-frequency dynamic components above the cutoff frequency reference, separating the low-frequency compensation state quantity; perform dynamic dimension transformation on the low-frequency compensation state quantity to generate the steady-state control torque residual.

[0007] Preferably, the specific process of inversely mapping the steady-state control torque residual to the body coordinate system to calculate and generate the real-time center of gravity offset vector, combined with the three-axis angular velocity sequence, is as follows: Perform quaternion attitude integration on the three-axis angular velocity sequence to generate a real-time rotation matrix representing the current body posture; construct the rigid body inverse dynamics Newton-Euler equations containing the body reference mass matrix and reference inertia tensor array; project the steady-state control torque residual into the rigid body inverse dynamics Newton-Euler equations through the real-time rotation matrix; perform cross-product matrix inverse decomposition on the projected torque term in the rigid body inverse dynamics Newton-Euler equations to extract the spatial position deviation of the mass distribution and generate the real-time center of gravity offset vector.

[0008] Preferably, the specific process of substituting the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor is as follows: extract the aircraft structural configuration parameters and construct a static configuration mapping matrix containing the spatial position vectors and torque conversion coefficients of each rotor; establish a three-dimensional spatial equilibrium state equation that constrains the net translational acceleration and net angular acceleration of the airframe to zero; substitute the real-time center of gravity offset vector into the three-dimensional spatial equilibrium state equation to solve the spatial compensation torque vector required to maintain the zero angular acceleration state; perform pseudo-inverse operation on the spatial compensation torque vector and the static configuration mapping matrix to allocate the thrust distribution components of each rotor branch and generate the steady-state thrust reference value of each rotor.

[0009] Preferably, the specific process for solving the cross-axis cross-coupling interference caused by asymmetric thrust distribution and converting it into a reverse yaw feedforward compensation component based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor is as follows: Input the steady-state thrust reference value of each rotor into the rotor aerodynamic anti-torsion characteristic function to generate the individual reverse drag torque corresponding to each independent rotor; Perform vector superposition operation on all individual reverse drag torques along the yaw axis of the fuselage to output the net yaw disturbance torque scalar; Import the net yaw disturbance torque scalar into the pitch-roll-yaw three-axis cross-coupling matrix to extract the yaw channel cross-interference term; Perform phase reversal and amplitude scaling processing on the yaw channel cross-interference term to generate the yaw feedforward compensation component.

[0010] Preferably, the specific process for calculating the remaining thrust margin of the steady-state thrust reference value and the extreme saturation thrust of each rotor, and mapping it to generate the dynamic response time constant of the corresponding operating point, is as follows: Extract the instantaneous maximum output power limit of the drive system, and generate the extreme saturation thrust boundary of each rotor by combining it with the current bus voltage; perform limit differential processing on the extreme saturation thrust boundary and the steady-state thrust reference value of each rotor to extract the remaining thrust margin sequence; construct a broadband response transfer function characterizing the electromagnetic inertia and aerodynamic damping coupling characteristics of the drive system; substitute the steady-state thrust reference value of each rotor into the broadband response transfer function to extract the step response delay characteristic of the current operating point; use the remaining thrust margin sequence to perform saturated nonlinear gain correction on the step response delay characteristic, and output the dynamic response time constant of the corresponding operating point.

[0011] Preferably, the remaining thrust margin and dynamic response time constant are used as joint constraints to perform high-frequency attitude adjustment proportional attenuation on rotors in low-margin and high-time-constant states. The specific process for constructing an asymmetric dynamic weight penalty matrix is ​​as follows: The remaining thrust margin and dynamic response time constant are orthogonally mapped in phase space to construct a two-dimensional joint state evaluation feature space; the out-of-bounds deviation of state points is extracted within the two-dimensional joint state evaluation feature space to locate the constrained rotor nodes in low-margin and high-time-constant states; the dynamic response time constant corresponding to the constrained rotor nodes is extracted and mapped to generate a high-frequency attitude command isolation cutoff frequency; the dynamic weight attenuation coefficient is calculated by combining the high-frequency attitude command isolation cutoff frequency and the out-of-bounds deviation of state points; the dynamic weight attenuation coefficients of all rotor nodes are updated along the diagonal of the identity matrix to generate an asymmetric dynamic weight penalty matrix.

[0012] Preferably, the specific process of generating the reconstructed control allocation matrix by performing a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the infrastructure configuration matrix is ​​as follows: Extract the infrastructure configuration matrix representing the physical installation position of the fixed rotor and the yaw torque coefficient; introduce the asymmetric dynamic weight penalty matrix into the quadratic objective function of control allocation, and configure the weights of the state deviation penalty terms; construct a weighted matrix equation containing the infrastructure configuration matrix and the asymmetric dynamic weight penalty matrix; perform singular value decomposition iterative solution on the weighted matrix equation to find the generalized inverse matrix solution that satisfies the minimum weighted norm constraint; extract the generalized inverse matrix solution to map and update the underlying control link distribution parameters, generating the reconstructed control allocation matrix.

[0013] Preferably, the specific process of superimposing the yaw feedforward compensation component onto the yaw control channel reference command and jointly reconstructing the control allocation matrix to convert the three-axis attitude control commands into independent duty cycle signals and sending them to each rotor actuator is as follows: Obtain the three-axis attitude control reference command sequence output by the outer-loop attitude controller; introduce the yaw feedforward compensation component into the yaw channel execution command reorganization in the three-axis attitude control reference command sequence to generate a corrected three-axis attitude control command sequence; jointly perform multi-dimensional vector space mapping operation with the corrected three-axis attitude control command sequence and the reconstructed control allocation matrix to parse and output the independent thrust demand vector of each rotor; extract the pre-calibrated rotor thrust and hardware duty cycle mapping feature table; substitute the independent thrust demand vector into the rotor thrust and hardware duty cycle mapping feature table to perform reverse lookup and hardware dead zone compensation calculation, converting it into a pulse width modulation duty cycle signal sequence and synchronously sending it to each rotor actuator.

[0014] A center-of-gravity sensing-driven aircraft dynamic control system, used to execute the aforementioned center-of-gravity sensing-driven aircraft dynamic control method, includes: a center-of-gravity decoupling module, used to synchronously acquire the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner-loop attitude controller; perform low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances; and extract the steady-state control torque residual characterizing the asymmetric load; combine the three-axis angular velocity sequence to back-map the steady-state control torque residual to the body coordinate system to calculate and generate a real-time center-of-gravity offset vector; and a feedforward compensation module, used to substitute the real-time center-of-gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor; and based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, solve the cross-axis cross-coupling caused by the asymmetric thrust distribution. The system performs several operations: 1) interference is processed and converted into a reverse yaw feedforward compensation component; 2) a dynamic weighting module is used to calculate the steady-state thrust baseline value and the remaining thrust margin of the extreme saturation thrust for each rotor, and map them to generate the dynamic response time constant of the corresponding operating point; 3) using the remaining thrust margin and dynamic response time constant as joint constraints, performs high-frequency attitude adjustment proportional attenuation on rotors with low margin and high time constant states, and constructs an asymmetric dynamic weight penalty matrix; 4) a control allocation module is used to perform a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the basic structure configuration matrix to generate a reconstructed control allocation matrix; 5) superimposing the yaw feedforward compensation component onto the yaw control channel baseline command, and jointly reconstructing the control allocation matrix to convert the three-axis attitude control commands into independent duty cycle signals and send them to each rotor actuator.

[0015] The technical effects and advantages of this invention are as follows: (1) In the technical solution described in this invention, the inertial characteristics and integral term state are acquired synchronously in the bottom control loop. Transient aerodynamic disturbances are isolated by low-pass filtering, and the steady-state control torque residual characterizing the asymmetric load is extracted. Then, the real-time center of gravity offset vector in the body coordinate system is calculated. On this basis, the steady-state thrust reference of each rotor required to maintain stability is accurately calculated by the dynamic balance equation, and the cross-axis cross-coupling interference hidden in the asymmetric thrust distribution is extracted and transformed into the reverse yaw feedforward compensation component. Thus, the problem of misjudgment of gust interference and structural resonance in the existing conventional control scheme can be overcome, the physical load offset can be accurately decoupled and extracted, and the unexpected yaw spin phenomenon caused by the center of gravity offset in the three-axis attitude adjustment can be effectively eliminated.

[0016] (2) In the technical solution described in this invention, based on the physical attenuation characteristics of the actuator under extreme conditions, the remaining thrust margin and dynamic response time constant corresponding to the steady-state thrust benchmark of each rotor are calculated. In this way, an asymmetric dynamic weighted penalty matrix is ​​constructed to proportionally attenuate the high-frequency attitude adjustment commands of the high-load rotor. Finally, the penalty matrix is ​​integrated into the weighted generalized inverse operation of the control allocation network to generate a reconstruction allocation mechanism, and the yaw feedforward compensation is synchronously superimposed on the bottom duty cycle command. Thus, the physical defects of overload saturation of the offset side rotor caused by the fixed thrust allocation strategy can be fundamentally solved. While ensuring the stability of the aircraft attitude, the high-frequency response burden of the rotor at the edge of saturation is significantly reduced, and the system's anti-saturation capability and cooperative control performance under complex flight conditions are improved.

[0017] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0018] Figure 1 This is a flowchart of a center-of-gravity sensing-driven dynamic control method for an aircraft according to the present invention. Figure 2 The nonlinear gain correction curve for dynamic response time constant provided in the embodiments of the present invention; Figure 3 This is a two-dimensional joint state evaluation and dynamic weight mapping surface diagram provided in an embodiment of the present invention; Figure 4 This is a flowchart of a center-of-gravity sensing-driven dynamic control system for aircraft according to the present invention. Detailed Implementation

[0019] This application provides a center-of-gravity sensing-driven dynamic control method and system for aircraft, which solves the problems of transaxial coupling runaway and offset side rotor overload saturation that easily occur when aircraft operate under asymmetric loads.

[0020] The overall approach of the scheme in this application is as follows: The physical center of gravity offset state is directly calculated using multi-source sensing data combined with low-pass filtering and inverse mapping. A yaw feedforward compensation channel is constructed by deriving the cross-coupling term caused by asymmetric thrust based on dynamic equilibrium. Simultaneously, the thrust saturation margin and response hysteresis characteristics of each rotor are evaluated, generating a dynamic weighted penalty matrix to limit the high-frequency actions of high-load rotors. Finally, the underlying thrust allocation logic is reconstructed based on weighted generalized inverse calculation, achieving accurate perception of physical load changes, underlying decoupling of cross-axis interference, and coordinated protection of actuator saturation margin.

[0021] Example 1; please refer to Figure 1 This invention provides a technical solution: a dynamic control method for an aircraft driven by center of gravity perception, comprising the following steps: S1, synchronously acquiring the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller; performing low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances; extracting the steady-state control torque residual characterizing the asymmetric load; combining the three-axis angular velocity sequence, mapping the steady-state control torque residual inversely to the body coordinate system to calculate and generate a real-time center of gravity offset vector; S2, substituting the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor; based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, solving the cross-axis cross-coupling interference caused by the asymmetric thrust distribution. S3. Calculate the remaining thrust margin of the steady-state thrust reference value and the extreme saturation thrust of each rotor, and map it to generate the dynamic response time constant of the corresponding operating point; use the remaining thrust margin and dynamic response time constant as joint constraints, and perform high-frequency attitude adjustment proportional decay on the rotor in the low margin and high time constant state to construct an asymmetric dynamic weight penalty matrix; S4. Perform a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the basic structure configuration matrix to generate a reconstructed control allocation matrix; superimpose the yaw feedforward compensation component onto the yaw control channel reference command, and jointly reconstruct the control allocation matrix to convert the three-axis attitude control command into an independent duty cycle signal and send it to each rotor actuator.

[0022] In this implementation, step S1 is used to perform state decoupling and center of gravity offset estimation. By reading the angular velocity data of the inertial measurement unit and the integral term data of the inner-loop attitude controller, low-pass filtering is used to remove the high-frequency signal components caused by transient gusts. The output sequence of the integral term of the inner-loop attitude controller refers to the feedback compensation value used in the attitude control algorithm to eliminate steady-state error over time. Under load conditions, the long-term steady-state accumulation of this value directly maps the asymmetric force characteristics of the aircraft's structure. The steady-state control torque residual is the absolute compensation torque required to resist the continuous center of gravity offset of the aircraft after filtering out transient aerodynamic disturbances. By directly superimposing it onto the yaw channel, it can actively cancel out the actual yaw displacement before the coupled spin motion produces a substantial yaw displacement.

[0023] Step S2 calculates the rotor thrust distribution necessary to maintain the current attitude at the dynamic level based on the extracted center of gravity offset vector and performs cross-axis decoupling. Since the center of gravity change requires each rotor to output uneven thrust to maintain the aircraft's horizontal position, the unbalanced aerodynamic drag torque corresponding to this asymmetric thrust distribution refers to the non-zero difference in reaction torque generated by each rotor due to different rotational speeds. Cross-axis cross-coupling interference refers to the unexpected rotational motion of the aircraft on the yaw axis caused by the aforementioned reaction torque difference when the aircraft adjusts its pitch or roll attitude to resist the center of gravity offset. This yaw feedforward compensation component is directly superimposed onto the yaw channel to construct a compensation feedforward path to offset the cross-axis disturbance torque.

[0024] Step S3 constrains and evaluates the physical saturation state and dynamic decay characteristics of the actuator and configures a control degradation strategy. Remaining thrust margin refers to the available difference between the current steady-state thrust output of the rotor and the maximum thrust that the motor hardware can provide. The dynamic response time constant characterizes the physical delay time required for the motor to perform a speed step change under the current heavy-load, high-speed condition. Since the heavy-load side rotor is in a near-full-load and slow-acceleration state, requiring it to perform high-frequency attitude fine-tuning easily leads to motor overload and control divergence. The asymmetric dynamic weight penalty matrix is ​​a state parameter matrix that reduces the control weight of a specific constrained rotor within a multi-dimensional control allocation framework. The system uses this matrix to proportionally reduce the high-frequency adjustment commands of the constrained rotor, allowing the low-load side rotor with sufficient margin to undertake the corresponding high-frequency attitude adjustment task.

[0025] Step S4 executes the reconstruction of the underlying control law and the synthesis and distribution of hardware drive signals. The weighted generalized inverse operation, in the case of control redundancy in multi-rotors, combines the basic structure configuration matrix (including physical installation positions) with the weight penalty matrix to solve for the optimal inverse matrix solution that conforms to the physical constraints of the actuators. The reconstructed control allocation matrix is ​​the thrust allocation rule set generated after the above calculation, directly replacing the original fixed static allocation ratio. Based on the reconstructed control allocation matrix, the system, in conjunction with yaw feedforward compensation commands, converts the three-axis attitude control requirements into independent duty cycle signals to adjust the conduction duty cycle of each electronic speed controller drive circuit, driving each rotor to operate collaboratively according to the asymmetric reconstruction strategy.

[0026] Specifically, the process of simultaneously acquiring the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller, performing low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances, and extracting the steady-state control torque residual characterizing the asymmetrical load is as follows: Read the integral term output sequences of the pitch and roll channels of the inner loop attitude controller; perform time-domain sliding window analysis on the three-axis angular velocity sequence to extract the high-frequency pulsating boundary frequency characterizing gust disturbances and structural resonance characteristics; set the high-frequency pulsating boundary frequency as the cutoff frequency reference of the adaptive digital filter; import the integral term output sequence into the adaptive digital filter to filter out high-frequency dynamic components above the cutoff frequency reference, separating the low-frequency compensation state quantity; perform dynamic dimension transformation on the low-frequency compensation state quantity to generate the steady-state control torque residual.

[0027] In this implementation, the pitch channel integral state sequence and roll channel integral state sequence output by the inner-loop attitude controller within the current control cycle are read and merged into a two-dimensional integral vector sequence. A time sliding window with a fixed sampling period is established, and a discrete Fourier transform is performed on the three-axis angular velocity sequence within each window to extract the frequency domain power spectral density. The discrete analytical formula for calculating the high-frequency pulsation boundary frequency is as follows: ;in, Indicates the high-frequency pulsation boundary frequency; This represents the total number of discrete frequency points within the power spectrum band; This represents the actual frequency value at the p-th discrete frequency point; This represents the power spectral density value of the triaxial angular velocity sequence at the p-th discrete frequency point. Directly mapped to the cutoff frequency reference of a first-order discrete adaptive digital filter, a dynamic filtering equation is established for the two-dimensional integral vector sequence and the low-frequency compensation state variables: ;in, This represents the low-frequency compensation state quantity of the current control cycle; This represents the low-frequency compensation state quantity of the previous control cycle; The input value represents the two-dimensional integral vector sequence of the current period; This represents the dynamic filtering smoothing coefficient. About The method for determining it is to calculate it based on the discrete-time transfer function, satisfying the physical constraint relationship. ,in This represents the fixed command sampling time interval of the aircraft control system. The inherent ratio of the static thrust output coefficient to the inner loop integral gain, pre-calibrated at the factory, of the aircraft's propulsion system is obtained and set as the rigid body moment mapping constant. The low-frequency compensation state quantity is multiplied by the rigid body moment mapping constant for dynamic dimension conversion, completing the three-dimensional extended conversion from the dimensionless control signal to the steady-state control torque residual with absolute mechanical and physical properties.

[0028] Specifically, the process of inversely mapping the steady-state control torque residual to the body coordinate system to calculate and generate the real-time center of gravity offset vector, combined with the three-axis angular velocity sequence, is as follows: Quaternion attitude integration is performed on the three-axis angular velocity sequence to generate a real-time rotation matrix representing the current body posture; a rigid body inverse dynamics Newton-Euler equation containing the body's reference mass matrix and reference inertia tensor array is constructed; the steady-state control torque residual is projected into the rigid body inverse dynamics Newton-Euler equation through the real-time rotation matrix; the projected torque term in the rigid body inverse dynamics Newton-Euler equation is decomposed by cross product matrix inverse decomposition to extract the spatial position deviation of the mass distribution, generating the real-time center of gravity offset vector.

[0029] In this implementation scheme, the quaternion matrix of the previous control cycle is extracted, and the attitude differential and integral operations are performed using the Picard approximation method with the current three-axis angular velocity sequence to obtain the quaternion parameters of the aircraft pose in the current cycle. A real-time rotation matrix representing the current aircraft pose is then constructed and expanded. ;in, This represents the real-time rotation matrix used for the transformation from the body coordinate system to the Earth navigation coordinate system. Represents the real part of the quaternion parameter of body position; This represents the imaginary vector component of the aircraft's pose quaternion parameter on the pitch axis. This represents the imaginary vector component of the body pose quaternion parameter on the roll axis. This represents the imaginary vector component of the aircraft's attitude quaternion parameter along the yaw axis. The reference total mass parameters of the aircraft under factory calibration are obtained, and a static distribution sequence including the Earth's gravitational acceleration vector is constructed. Since the steady-state control torque residual is entirely generated by the additional gravitational torque caused by the shift of the physical center of gravity, the rigid body inverse dynamics Newton-Euler equilibrium equations including the reference total mass parameters are established: ;in, This represents the steady-state control torque residual; This represents the spatial deviation of the three-dimensional physical mass distribution in the body coordinate system; Indicates the body's baseline total mass parameter; Represents the Earth's gravitational acceleration vector; Represents the transpose of the real-time rotation matrix; This indicates the cross product operation identifier for a three-dimensional vector. It performs an inverse decomposition operation on the cross product terms in the Newton-Euler equilibrium equations of rigid body inverse dynamics, transforming the gravity projection components in the body coordinate system into antisymmetric matrix operators, and generating a real-time center of gravity offset vector. ;in, This represents the real-time centroid offset vector generated after removing attitude distortion; This represents an operator that transforms an input three-dimensional column vector into a third-order antisymmetric matrix; This describes the process of obtaining the Moore-Ponros generalized inverse matrix from the extracted third-order antisymmetric matrix. By performing a generalized inverse operation on the gravity projection features, the spatial geometric projection interference caused by the aircraft's attitude tilt is removed, and the absolute center of gravity coordinate offset data in three-dimensional space is independently separated.

[0030] Specifically, the process of substituting the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor is as follows: extract the aircraft structural configuration parameters and construct a static configuration mapping matrix containing the spatial position vectors and torque conversion coefficients of each rotor; establish a three-dimensional spatial equilibrium state equation that constrains the net translational acceleration and net angular acceleration of the airframe to zero; substitute the real-time center of gravity offset vector into the three-dimensional spatial equilibrium state equation to solve the spatial compensation torque vector required to maintain the zero angular acceleration state; perform pseudo-inverse operation on the spatial compensation torque vector and the static configuration mapping matrix to allocate the thrust distribution components of each rotor branch and generate the steady-state thrust reference value of each rotor.

[0031] In this implementation scheme, the factory-set three-dimensional coordinate model parameters of the aircraft's physical frame are read, the spatial position vectors of each rotor's geometric center relative to the fuselage origin are extracted, and the torque conversion coefficients at corresponding speeds are obtained by combining the motor-propeller combination bench test, thus constructing a static configuration mapping matrix characterizing the spatial thrust distribution of the multi-rotor: ;in, This represents the static configuration mapping matrix; m represents the total number of rotors of the aircraft. This represents the projection component of the c-th rotor spatial position vector onto the vertical axis; This represents the projection component of the c-th rotor spatial position vector onto the horizontal axis; This represents the torque conversion coefficient corresponding to the c-th rotor. A static equilibrium equation is established for the constrained airframe in three-dimensional space, where both translational and angular accelerations are constant at zero. Since the system is only subjected to offset gravity and rotor thrust, the real-time center of gravity offset vector is substituted into this equation to solve for the spatial compensation torque vector under fully trimmed conditions. ;in, This represents the calculated spatial compensation torque vector; This represents an operator that transforms a spatial vector into a third-order skew-symmetric matrix; The magnitude scalar representing the vector of Earth's gravitational acceleration; This represents the unit direction vector vertically downwards in the body coordinate system. The vertical anti-gravity requirement scalar and the spatial compensation moment vector are combined into a generalized force requirement column vector. A weighted pseudo-inverse decomposition considering the frame structural strength is performed on the static configuration mapping matrix to solve and output the steady-state thrust baseline values ​​for each rotor. ;in, This represents the generated column vector of steady-state thrust reference values, which includes the thrust distribution components of each rotor branch. The transpose of the static configuration mapping matrix; This represents the diagonal matrix of structure weights. The determination method is based on the finite element modal analysis results of the frame arm stiffness. After normalizing the elastic deformation cutoff frequency of each branch arm, the matrix main diagonal position is assigned to achieve the physical safety constraint of preferentially distributing thrust to the high-strength side rotor.

[0032] Specifically, based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, the process of solving the cross-axis cross-coupling interference caused by the asymmetric thrust distribution and converting it into a reverse yaw feedforward compensation component is as follows: Input the steady-state thrust reference value of each rotor into the rotor aerodynamic anti-torsion characteristic function to generate the individual reverse drag torque corresponding to each independent rotor; perform vector superposition operation on all individual reverse drag torques along the yaw axis of the fuselage to output the net yaw disturbance torque scalar; import the net yaw disturbance torque scalar into the pitch-roll-yaw three-axis cross-coupling matrix to extract the yaw channel cross-interference term; perform phase reversal and amplitude scaling processing on the yaw channel cross-interference term to generate the yaw feedforward compensation component.

[0033] In this implementation scheme, based on the characteristics of aerodynamic viscous drag and induced downwash, the elements in the column vector containing the steady-state thrust reference value of each rotor are substituted one by one into the nonlinear rotor aerodynamic anti-torsion characteristic function to calculate the individual reverse drag torque generated by each independent rotor relative to the airframe when operating at the steady-state thrust reference value: ;in, This represents the single-unit reverse drag torque corresponding to the c-th rotor; This represents the inherent rotation direction identification factor of the c-th rotor, with a value of either positive or negative one. This represents the c-th scalar element in the column vector of steady-state thrust reference values ​​for each rotor. Indicates the laminar friction damping coefficient; This represents the nonlinear drag coefficient of the induced downwash airflow. A one-dimensional scalar superposition of the reverse drag moments of all individual units along the airframe dynamic yaw axis is performed to establish the net yaw disturbance moment scalar for the entire aircraft under an asymmetric thrust distribution environment: ;in, This represents the output net yaw disturbance moment scalar. A pitch-roll-yaw three-axis cross-coupling matrix reflecting the coupling interference between the airframe's asymmetric mass distribution and aerodynamic shape is constructed. The net yaw disturbance moment scalar is imported into this coupling system as an excitation source, and the yaw channel cross-interference term is independently extracted. ;in, This represents a spatial column vector containing the cross-interference factors of the three attitude axes; This represents the extracted yaw channel cross-interference term; This represents the pitch-roll-yaw three-axis cross-coupling matrix. The yaw channel cross-interference term undergoes in-phase processing to construct cancellation logic, and is multiplied by a pre-defined damping mapping gain to generate a yaw feedforward compensation component used to pre-eliminate coupled spin tendencies. ;in, This represents the final generated yaw feedforward compensation component; This represents the amplitude scaling factor. The method for determining the value is as follows: extract the proportional gain term of the inner loop yaw rate controller and the nominal value of the airframe yaw moment of inertia, calculate the quotient of the proportional gain term and the nominal value of the moment of inertia, and use it as a dimensionless mapping coefficient for direct configuration to accurately match the dynamic response bandwidth of the actuator control loop.

[0034] Specifically, the process of calculating the remaining thrust margin of the steady-state thrust reference value and the extreme saturation thrust of each rotor, and mapping it to generate the dynamic response time constant of the corresponding operating point is as follows: Extract the instantaneous maximum output power limit of the drive system, and generate the extreme saturation thrust boundary of each rotor by combining it with the current bus voltage; perform limit differential processing on the extreme saturation thrust boundary and the steady-state thrust reference value of each rotor to extract the remaining thrust margin sequence; construct a broadband response transfer function characterizing the electromagnetic inertia and aerodynamic damping coupling characteristics of the drive system; substitute the steady-state thrust reference value of each rotor into the broadband response transfer function to extract the step response delay characteristic of the current operating point; use the remaining thrust margin sequence to perform saturated nonlinear gain correction on the step response delay characteristic, and output the dynamic response time constant of the corresponding operating point.

[0035] In this implementation scheme, the current bus voltage fed back by the battery management system and the instantaneous maximum output power limit of the drive system calibrated at the underlying level of the motor ESC hardware are read. Based on the physical characteristic that the power of a brushless DC motor is proportional to the cube root of its static thrust, the extreme saturation thrust boundaries of each rotor are generated: ;in, This represents the extreme saturation tension boundary corresponding to the v-th rotor; v represents the rotor index. Indicates the overall electric propulsion conversion factor; Indicates the instantaneous maximum output power limit of the drive system; This represents the current bus voltage. By subtracting the steady-state thrust reference value corresponding to the same node from the extreme saturation thrust boundary, the remaining thrust margin characterizing the physical actuator thrust redundancy space is extracted. ;in, This represents the remaining thrust margin corresponding to the v-th rotor; This represents the v-th scalar in the extracted column vector of steady-state thrust reference values ​​for each rotor. To accurately reconstruct the hysteresis effect at high speeds, a broadband response transfer function integrating the coupling characteristics of electromagnetic coil inertia and blade aerodynamic damping is constructed. Substituting the steady-state thrust reference values ​​into the damping pole equation of this transfer function, the step response delay characteristic of the current operating point is analytically derived. ;in, This represents the step response delay characteristic at the current operating point; Represents the fundamental electromagnetic inertial constant; Indicates the air density in the current working environment; This represents the rotor aerodynamic damping coupling factor. When the thrust approaches the saturation limit, the motor response delay becomes nonlinearly amplified sharply. Therefore, a residual thrust margin is introduced to perform saturated nonlinear gain correction on the delay characteristic, outputting the final dynamic response time constant: ;in, This represents the dynamic response time constant at the operating point corresponding to the v-th rotor; Indicates the base gain with saturation penalty; Represents the nonlinear decay curvature coefficient; regarding The method for determining this is as follows: Step deceleration curves within the duty cycle range above 90% are extracted from rotor bench tests. The velocity residual data is then fitted using the natural logarithmic least squares method, and the optimal approximation slope is extracted as the curvature coefficient. (Refer to the relevant reference.) Figure 2 The horizontal axis in the figure represents the remaining thrust margin percentage (i.e., The vertical axis represents the corrected dynamic response time constant. .Depend on Figure 2 It can be seen that when the remaining thrust margin ratio is in a high safe range (e.g., greater than 60%), the dynamic response time constant remains stable at the baseline delay level. However, when the remaining thrust margin ratio decreases sharply and approaches zero (i.e., approaches the hardware saturation limit), the dynamic response time constant increases exponentially nonlinearly due to electromagnetic coil saturation and back EMF limitations. This curve fully reflects the hysteresis deterioration law of the actuator under extreme operating conditions, providing an objective physical boundary basis for subsequent weighting penalties.

[0036] Specifically, the remaining thrust margin and dynamic response time constant are used as joint constraints to perform high-frequency attitude adjustment proportional attenuation on rotors in low-margin and high-time-constant states. The specific process of constructing the asymmetric dynamic weight penalty matrix is ​​as follows: The remaining thrust margin and dynamic response time constant are orthogonally mapped in phase space to construct a two-dimensional joint state evaluation feature space; the out-of-bounds deviation of state points is extracted in the two-dimensional joint state evaluation feature space to locate the constrained rotor nodes in low-margin and high-time-constant states; the dynamic response time constant corresponding to the constrained rotor nodes is extracted and mapped to generate the high-frequency attitude command isolation cutoff frequency; the dynamic weight attenuation coefficient is calculated by combining the high-frequency attitude command isolation cutoff frequency and the out-of-bounds deviation of state points; the dynamic weight attenuation coefficients of all rotor nodes are arranged and updated along the diagonal of the identity matrix to generate the asymmetric dynamic weight penalty matrix.

[0037] In this implementation scheme, the extracted remaining thrust margin is converted into a normalized thrust load rate, and together with the normalized dynamic response time constant, it is used as orthogonal base coordinates to map the physical operating point of each rotor into a two-dimensional joint state evaluation feature space: ;in, This represents the normalized thrust load factor generated by the mapping; This represents the normalized response hysteresis rate generated by the mapping; This indicates the maximum allowable reference delay of the flight control system. To avoid control law oscillations caused by traditional hard threshold switching, a smooth Softplus nonlinear activation function is used to calculate the out-of-bounds deviation of the state point, achieving continuous and accurate positioning of rotors with low margins and high time constants. ;in, This represents the out-of-bounds deviation of the state point corresponding to the v-th rotor; Indicates load rate sensitive weight; Indicates the hysteresis-sensitive weight; Represents the joint security boundary bias scalar; regarding The determination method is as follows: based on the minimum phase margin constraint requirement for ensuring the absolute stability of the closed-loop system in frequency domain control theory, the critical stability envelope boundary value is set by inversely solving the Nyquist stability criterion. The original time constant is then frequency-domain broadened using the state point out-of-bounds deviation, mapping and generating the high-frequency attitude command isolation cutoff frequency that blocks high-frequency control signals. ;in, This represents the high-frequency attitude command isolation cutoff frequency corresponding to the v-th rotor. The dynamic weight attenuation coefficient of the control authority is calculated by combining the high-frequency attitude command isolation cutoff frequency with the inner-loop control reference frequency. ;in, This represents the dynamic weight attenuation coefficient corresponding to the v-th rotor; This represents the reference operating frequency of the inner loop control. The dynamic weight attenuation coefficients of all rotor nodes are extracted and distributed along the main diagonal of the identity matrix in rotor number order. The remaining off-diagonal elements are padded with zeros to construct an asymmetric dynamic weight penalty matrix for reshaping the control allocation space. ;in, This represents the asymmetric dynamic weight penalty matrix; The diagonal matrix construction operator is represented; m represents the total number of rotor nodes in the system. This matrix naturally weakens the weight of high-frequency attitude commands assigned to restricted rotors in the underlying generalized inverse operation by reducing the diagonal coefficient value corresponding to high-load rotors, thereby achieving high-frequency dynamic adaptive load reduction at the system level.

[0038] Specifically, the process of generating a reconstructed control allocation matrix by performing a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the infrastructure configuration matrix is ​​as follows: Extract the infrastructure configuration matrix representing the physical installation position of the fixed rotor and the yaw torque coefficient; introduce the asymmetric dynamic weight penalty matrix into the quadratic objective function of control allocation, configuring the weights of the state deviation penalty term; construct a weighted matrix equation containing the infrastructure configuration matrix and the asymmetric dynamic weight penalty matrix; perform singular value decomposition iterative solution on the weighted matrix equation to find the generalized inverse matrix solution satisfying the minimum weighted norm constraint; extract the generalized inverse matrix solution to map and update the underlying control link distribution parameters, generating the reconstructed control allocation matrix. (See also...) Figure 3 , Figure 3 In the figure, the orthogonal coordinate system on the bottom surface consists of the normalized thrust load rate and the normalized response hysteresis rate, respectively, with the Z-axis representing the dynamic weight attenuation coefficient of the output. The dark, flat area in the figure indicates that the system is within an absolutely safe envelope with sufficient margin and low latency, and the weight attenuation coefficient tends to 0. When the joint coordinate of the load rate and hysteresis rate crosses the boundary of the critical stability envelope, driven by the Softplus activation function and frequency domain broadening calculation, the surface rapidly bulges to form a nonlinear attenuation slope until the attenuation coefficient approaches 1 under extreme conditions. The shape distribution of this surface verifies that the algorithm can smoothly and continuously deprive high-frequency adjustment authority of high-load rotors, effectively avoiding the attitude oscillation problem caused by traditional hard threshold switching.

[0039] In this implementation scheme, the rotor physical installation coordinates and propeller aerodynamic yaw torque coefficients, which are fixed in the aircraft configuration database, are read to construct a static foundation matrix that maps spatial torque to each independent actuator, i.e., the basic structure configuration matrix. To establish an optimal allocation mechanism that balances attitude tracking accuracy and underlying dynamic saturation constraints, a quadratic objective function for control allocation is established. An asymmetric dynamic weight penalty matrix is ​​introduced into the control energy consumption term, and the weights of the state deviation penalty term are configured. This constructs the Heisenberg intermediate matrix in the weighted matrix equation reflecting the system's dynamic constraints. ;in, Represents the Heisenberg intermediate matrix; Represents the infrastructure configuration matrix; This represents the transpose of the infrastructure configuration matrix. This represents the asymmetric dynamic weight penalty matrix extracted in the previous process; This represents the weight matrix of the state deviation penalty term. Regarding... The method for determining the error tolerance is as follows: The background noise covariance matrix of the airborne inertial measurement unit under calibration conditions and the steady-state tracking error tolerance of the outer-loop control law are extracted. The joint reciprocal of these two factors is used as the diagonal element distribution, ensuring that the system prioritizes sacrificing the accuracy of the control channel with high fault tolerance when thrust is limited. Since the system matrix is ​​prone to local ill-conditioned characteristics after introducing the asymmetric penalty term, singular value decomposition is performed iteratively on the Heisenberg intermediate matrix. ;in, This represents the left orthogonal singular vector matrix obtained from the decomposition; This represents a diagonal matrix containing non-zero singular values; This represents the transpose of a right orthogonal singular vector matrix. Based on matrix theory satisfying the minimum weighted norm constraint, the generalized inverse matrix solution mapping is extracted and updated in the underlying control link to generate the reconstructed control allocation matrix: ;in, This represents the final generated reconstruction control allocation matrix; This represents the regular inverse of the singular value diagonal matrix. This reconstructed control allocation matrix directly reduces the response gain of high-temperature, low-margin rotor nodes at the computational physics level.

[0040] Specifically, the process of superimposing the yaw feedforward compensation component onto the yaw control channel reference command and jointly reconstructing the control allocation matrix to convert the three-axis attitude control commands into independent duty cycle signals and sending them to each rotor actuator is as follows: Obtain the three-axis attitude control reference command sequence output by the outer-loop attitude controller; introduce the yaw feedforward compensation component into the yaw channel execution command reorganization in the three-axis attitude control reference command sequence to generate a corrected three-axis attitude control command sequence; jointly perform multi-dimensional vector space mapping operations with the corrected three-axis attitude control command sequence and the reconstructed control allocation matrix to parse and output the independent thrust demand vector of each rotor; extract the pre-calibrated rotor thrust and hardware duty cycle mapping feature table; substitute the independent thrust demand vector into the rotor thrust and hardware duty cycle mapping feature table to perform reverse lookup and hardware dead-zone compensation calculation, converting it into a pulse width modulation duty cycle signal sequence and synchronously sending it to each rotor actuator.

[0041] In this implementation scheme, the three-axis attitude control reference command sequence generated after cascade PID calculation or sliding mode calculation is read from the outer loop attitude controller of the system. The reference command component corresponding to the yaw control channel is extracted, and the calculated reverse yaw feedforward compensation component is directly algebraically superimposed to generate the corrected three-axis attitude control command sequence. ;in, This represents the modified three-axis attitude control command sequence vector; This indicates the original roll attitude control command; This indicates the original pitch attitude control command; This indicates the original yaw attitude control command; This represents the yaw feedforward compensation component. The joint correction of the three-axis attitude control command sequence and the reconstructed control allocation matrix undergo multi-dimensional vector space mapping operations to analytically output the absolute thrust values ​​required by each rotor in the current cycle. ;in, This represents the independent thrust demand vector for each rotor as output by the analytical method. The rotor thrust-to-hardware duty cycle mapping characteristic table obtained from the motor bench calibration experiment is retrieved. A reverse lookup operation is performed on each element in the independent thrust demand vector, and hardware dead-zone compensation is calculated based on the physical start-up characteristics of the underlying electronic speed controller, transforming it into a pulse-width modulation duty cycle signal sequence. ;in, This represents the pulse width modulation duty cycle signal output to the d-th rotor actuator; d represents the sorting index variable of the bottom rotor channel. This represents the scalar value of the d-th element in the independent thrust demand vector; The nonlinear inverse interpolation addressing function represents the characteristic table of rotor thrust and hardware duty cycle mapping; This represents the hardware dead-time compensation value corresponding to the d-th rotor. (About...) The determination method is as follows: With the aircraft powered on and stationary, the electronic speed controller gradually increases the test pulse width from zero in minimum increments. High-frequency acquisition of motor phase current and rotor encoder signals is performed. The critical pulse width value corresponding to the first continuous steady-state rotational jump of the rotor is recorded as the hardware dead-zone compensation value for that execution channel. Finally, the generated pulse width modulation duty cycle signal sequence is synchronously sent to each actuator to complete the physical conversion and drive of electromechanical energy.

[0042] Example 2; please refer to Figure 4A center-of-gravity sensing-driven aircraft dynamic control system is used to execute a center-of-gravity sensing-driven aircraft dynamic control method described in the embodiments. The system includes: a center-of-gravity decoupling module, used to synchronously acquire the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner-loop attitude controller; perform low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances; and extract the steady-state control torque residual characterizing the asymmetric load. Combining the three-axis angular velocity sequence, the steady-state control torque residual is back-mapped to the body coordinate system to calculate a real-time center-of-gravity offset vector. A feedforward compensation module is used to substitute the real-time center-of-gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value for each rotor; and based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, the cross-axis cross-current caused by the asymmetric thrust distribution is calculated. The system identifies and converts cross-coupled interference into inverse yaw feedforward compensation components. A dynamic weighting module calculates the steady-state thrust baseline and residual thrust margin of each rotor, mapping them to generate the corresponding dynamic response time constant at the operating point. Using the residual thrust margin and dynamic response time constant as joint constraints, it performs high-frequency attitude adjustment proportional attenuation on rotors with low margin and high time constant states, constructing an asymmetric dynamic weight penalty matrix. A control allocation module performs a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the basic structure configuration matrix to generate a reconstructed control allocation matrix. The yaw feedforward compensation components are superimposed onto the yaw control channel baseline command, and the jointly reconstructed control allocation matrix converts the three-axis attitude control commands into independent duty cycle signals, which are then sent to each rotor actuator.

[0043] In this implementation scheme, the center of gravity decoupling module, as the front-end sensing and execution unit of the system, establishes a real-time data communication link with the airborne inertial measurement unit and the inner-loop attitude controller, synchronously acquiring the three-axis angular velocity sequence and the attitude integral term output sequence. Internally, this module invokes a digital filtering program to perform a low-pass filtering algorithm on the acquired integral term sequence, filtering out data fluctuations caused by high-frequency transient disturbances and separating the steady-state control torque residual. Subsequently, using the three-axis angular velocity sequence as the reference state boundary, the module performs coordinate transformation calculations on the separated steady-state control torque residual, mapping it inversely to the aircraft's dynamic coordinate system, calculating and outputting the real-time center of gravity offset vector in physical space.

[0044] The feedforward compensation module is connected to the center of gravity decoupling module. It receives the real-time center of gravity offset vector and substitutes it into the system's preset dynamic equilibrium equations. Through spatial statics calculations, it solves for the steady-state thrust reference value required for each rotor to maintain its current asymmetric state. After obtaining the steady-state thrust reference value for each rotor, this module extracts the corresponding anti-torsional torque characteristics to determine the unbalanced aerodynamic drag torque. Then, in a three-axis cross-coupled system, it solves for the yaw channel cross-interference term caused by this asymmetric thrust distribution. Finally, the module performs an inverse phase transformation operation on this interference term to generate a yaw feedforward compensation component with the opposite direction.

[0045] The dynamic weighting module is used for the low-level redistribution of system control authority. It retrieves the steady-state thrust baseline values ​​of each rotor from the feedforward compensation module, calculates the corresponding remaining thrust margin by combining them with the current extreme saturation thrust boundary of the system, and generates the dynamic response time constant corresponding to the current operating point of each rotor through a mapping program. This module uses the calculated remaining thrust margin and dynamic response time constant as two-dimensional joint constraints to locate constrained rotors with low margin and high time constant in the control matrix. For the matrix elements corresponding to such rotors, it executes a proportional decay algorithm for high-frequency attitude adjustment commands, generating and outputting an asymmetric dynamic weighting penalty matrix containing state constraint parameters.

[0046] The control allocation module, as the final command synthesis and distribution unit of the system, receives the asymmetric dynamic weight penalty matrix output by the dynamic weight module and the locally fixed infrastructure configuration matrix. It performs a weighted generalized inverse decomposition iterative operation on these two matrices to calculate the reconstructed control allocation matrix. Simultaneously, this module receives the three-axis attitude control reference commands issued from the system's outer loop and writes the yaw feedforward compensation components generated by the feedforward compensation module into the yaw control channel in an algebraic superposition form. Finally, this module, in conjunction with the reconstructed control allocation matrix, performs multi-dimensional spatial mapping calculations on the corrected three-axis attitude control commands, converting the physical requirements of each channel into independent duty cycle signals, and synchronously sends them to the underlying rotor actuators.

[0047] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for dynamic control of an aircraft driven by center of gravity sensing, characterized in that, Includes the following steps: S1. Simultaneously acquire the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller. Perform low-pass filtering on the integral term output sequence to isolate high-frequency transient disturbances and extract the steady-state control torque residual characterizing the asymmetrical load. Combine the three-axis angular velocity sequence and back-map the steady-state control torque residual to the body coordinate system to calculate and generate the real-time center of gravity offset vector. S2. Substitute the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor; based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, solve the cross-axis cross-coupling interference caused by the asymmetric thrust distribution, and convert it into the reverse yaw feedforward compensation component. S3. Calculate the remaining thrust margin of the steady-state thrust benchmark value and the extreme saturation thrust of each rotor, and map it to generate the dynamic response time constant of the corresponding operating point; use the remaining thrust margin and dynamic response time constant as joint constraints, perform high-frequency attitude adjustment proportional decay on the rotor in the low margin and high time constant state, and construct an asymmetric dynamic weight penalty matrix. S4. Perform a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the infrastructure configuration matrix to generate a reconstructed control allocation matrix. The yaw feedforward compensation component is superimposed on the yaw control channel reference command, and the joint reconstruction of the control allocation matrix converts the three-axis attitude control commands into independent duty cycle signals and sends them to each rotor actuator.

2. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 1, characterized in that: The specific process of simultaneously acquiring the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller, low-pass filtering the integral term output sequence to isolate high-frequency transient disturbances, and extracting the steady-state control torque residual characterizing the asymmetrical load is as follows: Read the integral output sequences of the pitch and roll channels of the inner loop attitude controller; A time-domain sliding window analysis was performed on the triaxial angular velocity sequence to extract the high-frequency pulsating boundary frequencies that characterize the gust disturbance and structural resonance. Set the high-frequency pulsation boundary frequency as the cutoff frequency reference for the adaptive digital filter; The integral term output sequence is imported into an adaptive digital filter to filter out high-frequency dynamic components above the cutoff frequency reference and separate the low-frequency compensation state quantity. Perform dynamic dimension transformation on the low-frequency compensation state variables to generate steady-state control torque residuals.

3. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 2, characterized in that: The specific process of combining the three-axis angular velocity sequence to back-map the steady-state control torque residual to the body coordinate system and calculate the real-time center of gravity offset vector is as follows: Perform quaternion attitude integration on the three-axis angular velocity sequence to generate a real-time rotation matrix representing the current body posture; Construct the Newton-Euler equations for rigid body inverse dynamics, which include the body's reference mass matrix and reference inertia tensor array; The steady-state control torque residual projection is introduced into the Newton-Euler equations of rigid body inverse dynamics by using a real-time rotation matrix. The projected torque term in the Newton-Euler equations of rigid body inverse dynamics is decomposed by cross product matrix inverse decomposition to extract the spatial position deviation of mass distribution and generate a real-time center of gravity offset vector.

4. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 1, characterized in that: The specific process of substituting the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor is as follows: Extract the structural configuration parameters of the aircraft and construct a static configuration mapping matrix that includes the spatial position vectors and torque conversion coefficients of each rotor. Establish the three-dimensional spatial equilibrium equations for the constrained body, where the net translational acceleration and net angular acceleration are both zero. Substitute the real-time center of gravity offset vector into the three-dimensional spatial equilibrium equation to calculate the spatial compensation torque vector required to maintain the zero angular acceleration state. The pseudo-inverse operation is performed on the spatial compensation torque vector and the static configuration mapping matrix to allocate the thrust distribution components of each rotor branch and generate the steady-state thrust reference value of each rotor.

5. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 4, characterized in that: Based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, the specific process of solving the cross-axis cross-coupling interference caused by the asymmetric thrust distribution and converting it into the reverse yaw feedforward compensation component is as follows: Input the steady-state thrust reference value of each rotor into the rotor aerodynamic anti-torsion characteristic function to map and generate the single-unit reverse drag torque corresponding to each independent rotor. A vector superposition operation is performed on all individual unit reverse drag moments along the yaw axis of the airframe to output a net yaw disturbance torque scalar. The net yaw disturbance moment scalar is imported into the pitch-roll-yaw three-axis cross coupling matrix, and the yaw channel cross interference term is extracted. Phase reversal and amplitude scaling processing are performed on the yaw channel cross interference term to generate yaw feedforward compensation components.

6. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 1, characterized in that: The specific process for calculating the remaining thrust margin of the steady-state thrust reference value and the extreme saturation thrust of each rotor, and mapping it to generate the dynamic response time constant of the corresponding operating point is as follows: Extract the instantaneous maximum output power limit of the drive system, and combine it with the current bus voltage to generate the extreme saturation thrust boundary of each rotor. Amplitude limiting differential processing is performed on the extreme value saturation thrust boundary and the steady-state thrust reference value of each rotor to extract the remaining thrust margin sequence; Construct a broadband response transfer function characterizing the electromagnetic inertia and aerodynamic damping coupling characteristics of the drive system; Substitute the steady-state thrust reference value of each rotor into the broadband response transfer function to extract the step response delay characteristic of the current operating point. The residual thrust margin sequence is used to perform saturated nonlinear gain correction on the step response delay characteristic, and the dynamic response time constant corresponding to the operating point is output.

7. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 6, characterized in that: Using the remaining thrust margin and dynamic response time constant as joint constraints, the rotor in a low-margin, high-time-constant state is subjected to high-frequency attitude adjustment proportional decay. The specific process of constructing the asymmetric dynamic weight penalty matrix is ​​as follows: The remaining thrust margin is orthogonally mapped to the dynamic response time constant in phase space to construct a two-dimensional joint state assessment feature space. Extract the out-of-bounds deviation of state points in the two-dimensional joint state assessment feature space to locate the confined rotor node with low margin and high time constant. Extract the dynamic response time constant corresponding to the constrained rotor node and map it to generate the high-frequency attitude command isolation cutoff frequency; The dynamic weight attenuation coefficient is calculated by combining the isolation cutoff frequency of high-frequency attitude commands with the deviation of state points from the boundary. The dynamic weight decay coefficients of all rotor nodes are updated by arranging them along the diagonal of the identity matrix, generating an asymmetric dynamic weight penalty matrix.

8. The method for dynamic control of an aircraft driven by center of gravity sensing according to claim 1, characterized in that: The specific process of generating the reconstructed control allocation matrix by performing a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the infrastructure configuration matrix is ​​as follows: Extract the basic structural configuration matrix that characterizes the physical installation position of the fixed rotor and the yaw torque coefficient of the aircraft. An asymmetric dynamic weight penalty matrix is ​​introduced into the quadratic objective function of control allocation to configure the weight of the state deviation penalty term. Construct a weighted matrix equation that includes the infrastructure configuration matrix and the asymmetric dynamic weight penalty matrix; Perform singular value decomposition iterative solution on the weighted matrix equation to find the generalized inverse matrix solution that satisfies the minimum weighted norm constraint; Extract the generalized inverse matrix, demap it to update the underlying control link distribution parameters, and generate a reconstructed control allocation matrix.

9. A method for dynamic control of an aircraft driven by center of gravity sensing according to claim 8, characterized in that: The specific process of superimposing the yaw feedforward compensation component onto the yaw control channel reference command and jointly reconstructing the control allocation matrix to convert the three-axis attitude control commands into independent duty cycle signals and sending them to each rotor actuator is as follows: Obtain the three-axis attitude control reference command sequence output by the outer loop attitude controller; The yaw feedforward compensation component is introduced into the yaw channel execution command reorganization in the three-axis attitude control reference command sequence to generate a corrected three-axis attitude control command sequence. The joint correction of the three-axis attitude control command sequence and the reconstruction of the control allocation matrix are performed to perform multi-dimensional vector space mapping operations, and the independent thrust demand vector of each rotor is output analytically. Extract the pre-calibrated rotor thrust and hardware duty cycle mapping feature table; The independent thrust demand vector is substituted into the rotor thrust and hardware duty cycle mapping feature table to perform reverse lookup and hardware dead zone compensation calculation, which is then converted into a pulse width modulation duty cycle signal sequence and synchronously sent to each rotor actuator.

10. A center-of-gravity sensing-driven aircraft dynamic control system, used to execute the center-of-gravity sensing-driven aircraft dynamic control method according to any one of claims 1-9, characterized in that, include: The center of gravity decoupling module is used to synchronously acquire the three-axis angular velocity sequence of the inertial measurement unit and the integral term output sequence of the inner loop attitude controller. The integral term output sequence is low-pass filtered to isolate high-frequency transient disturbances and extract the steady-state control torque residual characterizing the asymmetrical load. Combined with the three-axis angular velocity sequence, the steady-state control torque residual is back-mapped to the body coordinate system to calculate and generate the real-time center of gravity offset vector. The feedforward compensation module is used to substitute the real-time center of gravity offset vector into the dynamic equilibrium equation to calculate the steady-state thrust reference value of each rotor; based on the unbalanced aerodynamic drag torque corresponding to the steady-state thrust reference value of each rotor, it solves the cross-axis cross-coupling interference caused by the asymmetric thrust distribution and converts it into the reverse yaw feedforward compensation component. The dynamic weight module is used to calculate the remaining thrust margin of the steady-state thrust benchmark value and the extreme saturation thrust of each rotor, and to map and generate the dynamic response time constant of the corresponding operating point. The remaining thrust margin and dynamic response time constant are used as joint constraints to perform high-frequency attitude adjustment proportional decay on rotors with low margin and high time constant, and to construct an asymmetric dynamic weight penalty matrix. The control allocation module is used to perform a weighted generalized inverse operation on the asymmetric dynamic weight penalty matrix and the basic structure configuration matrix to generate a reconstructed control allocation matrix; it also superimposes the yaw feedforward compensation component onto the yaw control channel reference command, and jointly reconstructs the control allocation matrix to convert the three-axis attitude control commands into independent duty cycle signals and send them to each rotor actuator.

Images