A center of gravity driven aircraft precise attitude regulation method and system

Abstract

The application discloses a center-of-gravity driving aircraft precise attitude regulation method and system, and particularly relates to the field of aircraft attitude control technology, and is used for solving the problem of high operation threshold and easy loss of control of man-machine coupling in a complex environment. First, cabin pressure distribution data is collected and frequency domain decoupling is performed, and a real low-frequency intention driving vector and an associated high-frequency jitter feature are accurately separated. Then, the intention is mapped to a flow field coordinate system that fuses terrain and airflow to perform aerodynamic interference impedance analysis, generate an aerodynamic impedance compensation matrix, and avoid collision risks. The high-frequency jitter and the expected body angular response matching resonance band are extracted, a dynamic anti-oscillation damping coefficient is output, and resonance is suppressed. Finally, multi-dimensional parameters are input into a nonlinear control solver to perform joint optimization and generate a thrust distribution sequence for execution, thereby constructing a regulation closed loop from intention decoupling to physical execution, and ensuring extreme smoothness and flight safety in an extremely simple operation mode.

Description

Technical Field

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

[0002] With the rapid development of the low-altitude economy and the gradual implementation of urban air mobility concepts, new types of manned flight equipment, such as personal vertical takeoff and landing (VTOL) aircraft, are rapidly entering the mass consumer market. Flight equipment targeting ordinary consumers places extremely high demands on ease of operation. Exploring a simplified flight operation mode that establishes a direct mechanical mapping between the dynamic shift of the occupant's physical center of gravity and the aircraft's spatial attitude response, thereby avoiding complex command conversions and reducing pilot learning costs, has become an important development direction for the large-scale commercialization of aircraft.

[0003] Most existing aircraft attitude control methods rely on physical joysticks, touchscreen buttons, or external handheld remote control devices for command input. These control methods suffer from a steep learning curve, requiring pilots to undergo extensive muscle memory training to translate visual information into precise hand-foot micro-operations. Furthermore, in the event of sudden turbulence or emergencies, traditional physical joystick controls necessitate complex spatial and command conversions by the brain, easily leading to delayed emergency response and potentially fatal misoperations. In addition, some existing solutions attempting motion-sensing control are highly susceptible to involuntary muscle tremors caused by turbulence or crew tension in the complex aerodynamic environment of high altitudes, potentially triggering severe human-machine coupling oscillations, resulting in aircraft attitude instability or even crashes. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for precise attitude control of aircraft driven by the center of gravity, thus solving the problems mentioned above.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a gravity-driven precise attitude control method for an aircraft, comprising the following steps: S1, acquiring the time-series matrix of pressure distribution between the aircraft cockpit floor and the seat contact surface; performing spatiotemporal gradient analysis on the pressure distribution time-series matrix to extract the three-dimensional coordinates of the transient pressure center of the human body; and performing frequency domain decoupling on the three-dimensional coordinates of the transient pressure center of the human body to separate the low-frequency intention driving vector and the accompanying high-frequency jitter feature packet; S2, mapping the low-frequency intention driving vector to the local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and airflow particle motion vectors acquired by the airborne sensing device; and substituting the body dynamics envelope boundary into the local flow field coordinate system to perform aerodynamic control. S3. Dynamic interference impedance analysis is performed to generate an aerodynamic impedance compensation matrix that includes spatial anti-disturbance margin; S4. The time-series amplitude sequence within the associated high-frequency jitter feature package is extracted, and the overlap feature extraction is performed on the expected body angle response frequency mapped by the time-series amplitude sequence and the aerodynamic impedance compensation matrix to match the human-machine coupling resonance band. The peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band are analyzed; S5. The low-frequency intentional drive vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient are input into the nonlinear control solver to perform joint optimization in the state space, generate the multi-rotor thrust distribution sequence within the control cycle, and convert the multi-rotor thrust distribution sequence into a pulse width modulation signal and send it to the underlying power mechanism.

[0006] Furthermore, the specific process of extracting the three-dimensional coordinates of the human transient pressure center by performing spatiotemporal gradient analysis on the pressure distribution time series matrix of the aircraft cockpit floor and seat contact surface is as follows: The discrete force signal sequence of the flexible piezoresistive array built into the cockpit floor and seat support surface is obtained; spatial surface interpolation is performed on the discrete force signal sequence to generate a continuous pressure topology surface; multi-channel clock synchronous sampling is performed on the continuous pressure topology surface to construct the pressure distribution time series matrix; second-order spatial gradient feature extraction is performed along the three axes of the cockpit space in the pressure distribution time series matrix to locate local pressure extremum clusters; spatial torque balance equivalent transformation is performed on the local pressure extremum clusters to output the three-dimensional coordinates of the human transient pressure center.

[0007] Furthermore, the specific process of performing frequency domain decoupling and separation of the low-frequency intention driving vector and the accompanying high-frequency jitter feature packet from the three-dimensional coordinates of the human transient pressure center is as follows: The three-dimensional coordinates of the human transient pressure center are arranged and reconstructed along the time axis to generate a motion trajectory sequence. Wavelet packet decomposition is performed on the motion trajectory sequence to extract multi-scale time-frequency domain feature coefficients. Based on the upper limit of the active force exertion frequency of human biomechanics, a cutoff frequency band splitting node is established. Low-frequency approximation coefficients below the cutoff frequency band splitting node are extracted and spatial vector reconstruction is performed to output the low-frequency intention driving vector. High-frequency detail coefficients above the cutoff frequency band splitting node are extracted and envelope detection convergence is performed to generate the accompanying high-frequency jitter feature packet.

[0008] Furthermore, the specific process of mapping the low-frequency intention-driven vector to the local flow field coordinate system constructed by fusing the 3D terrain point cloud and the airflow particle motion vector collected by the airborne sensing device is as follows: Voxelization spatial downsampling is performed on the 3D terrain point cloud to establish a static terrain boundary matrix; the airflow particle motion vector is projected onto the non-occupied voxels of the static terrain boundary matrix to assign dynamic wind field attributes, thus constructing a local flow field coordinate system; the real-time pose angle parameters of the aircraft relative to the local flow field coordinate system are extracted to construct a spatial attitude transformation matrix; based on the spatial attitude transformation matrix, coordinate system alignment analysis is performed on the low-frequency intention-driven vector under the local reference of the cockpit to complete the spatial projection mapping to the local flow field coordinate system.

[0009] Furthermore, the specific process of performing aerodynamic interference impedance analysis in the local flow field coordinate system by substituting the body dynamics envelope boundary into the aerodynamic impedance compensation matrix containing spatial disturbance rejection margin is as follows: Extending along the direction of the low-frequency intentional drive vector after spatial projection mapping, extract the directional airflow distribution gradient and terrain approximation distance in the local flow field coordinate system; performing joint boundary extremum determination on the directional airflow distribution gradient, terrain approximation distance, and the maximum lift coefficient and ultimate tilt angle in the body dynamics envelope boundary to extract the boundary risk factor; substituting the boundary risk factor into the preset aerodynamic impedance potential field model to deduce the normal repulsive force feedback tensor against the directional airflow distribution gradient and terrain approximation distance; performing dynamic vector synthesis on the normal repulsive force feedback tensor and the low-frequency intentional drive vector after spatial projection mapping to output the aerodynamic impedance compensation matrix containing spatial disturbance rejection margin.

[0010] Furthermore, the specific process of extracting the time-series amplitude sequence within the associated high-frequency jitter feature package and performing overlap feature extraction and matching between the time-series amplitude sequence and the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to obtain the human-machine coupling resonance band is as follows: The instantaneous energy envelope of each frequency band within the associated high-frequency jitter feature package is extracted to construct the time-series amplitude sequence. A dynamic state-space transformation is performed on the aerodynamic impedance compensation matrix to extract the expected body angle response frequency under the current flow field load. Cross-power spectral density estimation is performed on the time-series amplitude sequence and the expected body angle response frequency to generate a frequency domain coherent feature spectrum. The extreme value interval of the coherence coefficient is traced in the frequency domain coherent feature spectrum to define the overlapping frequency band. Dynamic closure constraints are applied to the upper and lower frequency boundaries of the overlapping frequency band to output the human-machine coupling resonance band.

[0011] Furthermore, the specific process of analyzing the peak intensity and duration period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonant band is as follows: within the human-machine coupling resonant band, the peak intensity of the extreme value feature is analyzed and quantized by resonant energy spectral density aggregation. The duration period is extracted by tracing the coherent feature residence time span of the human-machine coupling resonant band along the time domain dimension. Nonlinear gain state matching is performed within the input damping adaptive attenuation mapping surface of the peak intensity and duration period. The gain adjustment extreme value of the output of the damping adaptive attenuation mapping surface is extracted and determined as the dynamic anti-oscillation attenuation coefficient.

[0012] Furthermore, the specific process of inputting the low-frequency intention driving vector, aerodynamic impedance compensation matrix, and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint state-space optimization and generate the multirotor thrust allocation sequence within the control cycle is as follows: The low-frequency intention driving vector is set as the target tracking reference state of the nonlinear control solver; the aerodynamic impedance compensation matrix is ​​transformed into a state constraint boundary and fused into the rolling optimization objective function of the nonlinear control solver; the dynamic anti-oscillation attenuation coefficient is reconstructed as the dynamic penalty weight matrix of the nonlinear control solver at the control increment level; within the rolling optimization objective function, multivariable dynamic state trajectory rolling deduction optimization of the finite-time domain control variable sequence is performed based on the rigid body dynamics model; the optimal control input matrix after the deduction optimization convergence is extracted, and the dynamic allocation solution is performed to output the multirotor thrust allocation sequence within the control cycle.

[0013] A center-of-gravity driven precision attitude control system for aircraft, used to execute the aforementioned center-of-gravity driven precision attitude control method for aircraft, includes: an intent decoupling module, used to acquire the pressure distribution time-series matrix of the aircraft cockpit floor and seat contact surface, perform spatiotemporal gradient analysis on the pressure distribution time-series matrix to extract the three-dimensional coordinates of the human transient pressure center, and perform frequency domain decoupling on the three-dimensional coordinates of the human transient pressure center to separate the low-frequency intent driving vector and the accompanying high-frequency jitter feature packet; and an aerodynamic anti-interference module, used to map the low-frequency intent driving vector to a local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and airflow particle motion vectors acquired by the airborne sensing equipment, and substitute it into the body dynamics envelope boundary to perform aerodynamic anti-interference within the local flow field coordinate system. Interference impedance analysis generates an aerodynamic impedance compensation matrix containing spatial anti-disturbance margin; the oscillation suppression module extracts the time-series amplitude sequence within the associated high-frequency jitter feature package, performs overlapping feature extraction and matching of the time-series amplitude sequence with the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to identify the human-machine coupling resonance band, and analyzes the peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band; the thrust allocation module inputs the low-frequency intentional drive vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint state-space optimization, generates a multi-rotor thrust allocation sequence within the control cycle, and converts the multi-rotor thrust allocation sequence into a pulse width modulation signal and sends it to the underlying power mechanism.

[0014] The present invention has the following beneficial effects: (1) A center-of-gravity driven method for precise attitude control of an aircraft. By acquiring the cockpit pressure distribution time-series matrix and performing frequency domain decoupling, the method accurately separates the occupant's true low-frequency intention drive vector and the accompanying high-frequency jitter feature packet. Subsequently, the intention is mapped to a local flow field coordinate system that integrates terrain and airflow to perform aerodynamic interference impedance analysis and generate an aerodynamic impedance compensation matrix. This design effectively overcomes the shortcomings of traditional somatosensory control, which is easily interfered with by irrational human movements. It not only filters out physiological tremors caused by occupant tension, but also avoids aerodynamic stall and collision risks in complex flow field environments in advance, significantly improving the accuracy of the aircraft's bottom-level perception and physical boundary safety in complex low-altitude environments.

[0015] (2) A center-of-gravity driven precision attitude control system for aircraft further extracts high-frequency jitter characteristics and matches the expected body angle response frequency to determine the human-machine coupling resonance band, outputs a dynamic anti-oscillation attenuation coefficient, and combines the intention vector and impedance matrix input nonlinear control solver to perform joint optimization in the state space. This design approaches the problem from the physical dimension of biomechanics and aerodynamics coupling, effectively suppressing the dangerous resonance phenomenon caused by crew stress swaying, eliminating the hysteresis of the underlying rotor physical response, realizing a smooth and advanced prediction fit between the aircraft attitude and the crew's real piloting intention, and ensuring the ultimate stability of the minimalist control mode in three-dimensional flight space.

[0016] 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

[0017] Figure 1 This is a flowchart of a center-of-gravity driven precise attitude control method for an aircraft according to the present invention.

[0018] Figure 2 A flowchart for human intent analysis and frequency domain decoupling.

[0019] Figure 3 This is a flowchart of the local flow field coordinate system mapping and aerodynamic impedance derivation.

[0020] Figure 4 This is a flowchart of the state-space optimization process for a nonlinear control solver.

[0021] Figure 5 This is a flowchart of a center-of-gravity driven precision attitude control system for an aircraft according to the present invention. Detailed Implementation

[0022] This application provides a center-of-gravity driven method and system for precise attitude control of aircraft, which solves the problems of high control threshold, slow emergency response, and easy loss of control of human-machine coupling in complex aerodynamic environments in existing aircraft.

[0023] The overall approach of the scheme in this application is as follows: First, cabin pressure distribution data is processed by spatiotemporal gradient analysis and frequency domain decoupling to accurately separate the occupant's actual low-frequency driving intentions from high-frequency physiological vibration noise. Second, the actual driving intentions are substituted into the flow field coordinate system combined with terrain and airflow motion to perform impedance analysis, generating an aerodynamic compensation state that includes physical disturbance rejection margin. Next, the overlapping features of high-frequency physiological vibrations and expected aircraft response are extracted to lock the human-machine coupling resonance band and output the anti-oscillation attenuation coefficient. Finally, the actual driving intentions, aerodynamic impedance compensation, and anti-oscillation attenuation coefficient are jointly input into the control solver for dynamic state optimization, outputting the physical thrust distribution command of the underlying power mechanism, thereby completing the closed-loop physical control from the human body's physical center of gravity deflection to the stable flight of the aircraft.

[0024] Example 1; please refer to Figure 1 This invention provides a technical solution: a method for precise attitude control of a center-of-gravity driven aircraft, comprising the following steps: S1, acquiring the pressure distribution time-series matrix of the contact surface between the aircraft cockpit floor and the seat; performing spatiotemporal gradient analysis on the pressure distribution time-series matrix to extract the three-dimensional coordinates of the transient pressure center of the human body; and performing frequency domain decoupling on the three-dimensional coordinates of the transient pressure center of the human body to separate the low-frequency intention driving vector and the accompanying high-frequency jitter feature packet; S2, mapping the low-frequency intention driving vector to the local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and the airflow particle motion vector acquired by the airborne sensing device; and substituting the body dynamics envelope boundary to perform aerodynamic interference within the local flow field coordinate system. S3. Anti-interference analysis is performed to generate an aerodynamic impedance compensation matrix containing spatial anti-interference margin; S4. The time-series amplitude sequence within the associated high-frequency jitter feature package is extracted, and the overlap feature extraction is performed on the expected body angle response frequency mapped by the time-series amplitude sequence and the aerodynamic impedance compensation matrix to match the human-machine coupling resonance band. The peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band are analyzed; S5. The low-frequency intentional drive vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient are input into the nonlinear control solver to perform joint optimization in the state space, generate the multi-rotor thrust allocation sequence within the control cycle, and convert the multi-rotor thrust allocation sequence into a pulse width modulation signal and send it to the underlying power mechanism.

[0025] In this implementation scheme, step S1 is mainly used to accurately capture the occupant's instinctive changes in driving posture and eliminate physiological interference. The pressure distribution time-series matrix refers to the set of pressure value changes collected by sensor arrays distributed on the cockpit floor and seats over a continuous time period; spatiotemporal gradient analysis refers to accurately locating the spatial geometric center of the overall force on the human body by calculating the rate and direction of pressure changes at each measurement point in the physical space; frequency domain decoupling refers to using digital signal processing to separate and decompose complex coordinate motion signals according to frequency. Through this step, the system can accurately obtain the low-frequency intention drive vector representing the occupant's actual driving direction and accurately classify the involuntary muscle tremors caused by high-altitude tension or turbulence into accompanying high-frequency jitter feature packets, thereby preventing the aircraft from becoming unstable due to receiving unconscious swaying commands from the human body.

[0026] Step S2 primarily involves placing the occupant's center of gravity and piloting intentions within a realistic external physical environment for safety boundary verification. The local flow field coordinate system refers to a three-dimensional physical environment model surrounding the aircraft, constructed by integrating terrain obstacle point clouds from airborne radar scans and external wind speed and direction data. The airframe dynamics envelope boundary represents the aircraft's physical and mechanical limits, such as the maximum tilt angle and maximum wind resistance level. Aerodynamic interference impedance analysis calculates the amount of external airflow resistance or terrain proximity risks the aircraft would encounter if it flew entirely in the direction of the occupant's center of gravity offset. Through this step, the system can anticipate external environmental hazards and proactively incorporate spatial anti-disturbance margins to counteract wind drift and prevent collisions into the occupant's original intention commands, generating an aerodynamic impedance compensation matrix to ensure the aircraft's absolute physical safety in complex low-altitude flow field environments.

[0027] Step S3 is primarily used to break the dangerous cycle of human-machine resonance caused by the superposition of occupant tension and swaying with changes in aircraft attitude. The expected aircraft angular response frequency refers to the frequency at which the aircraft's physical structure will exhibit attitude swaying after receiving simulated commands with aerodynamic compensation. Overlap feature extraction involves comparing the high-frequency shaking frequency of the occupant's body with the aircraft's swaying frequency to identify the dangerous frequency band where they overlap. The human-machine coupling resonance band refers to the specific frequency region where involuntary occupant swaying and aircraft swaying mutually amplify and excite each other. Through this step, the system can detect the occupant's physiological tension and the aircraft's resonance physical risk in real time. Once a resonance trend is detected, a dynamic anti-oscillation attenuation coefficient is immediately calculated, providing adaptive damping parameters for the subsequent power system to passivate the aircraft's physical response, ensuring smooth flight and occupant comfort in the minimalist center-of-gravity control mode.

[0028] Step S4 primarily coordinates human intention signals and environmental impedance compensation parameters to complete the final physical dynamic closed-loop distribution output of the aircraft. The nonlinear control solver is an algorithm unit specifically designed to handle the complex, multivariable, nonlinear aerodynamic model of multirotor aircraft. State-space joint optimization calculates the optimal motor speed control strategy under multiple physical constraints, including satisfying the occupant's physical center of gravity deflection direction, resisting complex external airflow interference, and suppressing human-machine coupling resonance. The pulse width modulation signal is the fundamental electrical signal that directly drives the underlying rotor motor ESC. Through this step, the system transforms the comprehensive state, after full boundary verification and environmental compensation, into a sequence of specific thrust values ​​allocated to each rotor motor at the physical level. This completely eliminates the hysteresis of the physical response in traditional rotor systems, achieving advanced, smooth, and safe underlying physical control of the aircraft's spatial attitude in response to the occupant's true center of gravity intention.

[0029] Please see Figure 2 Specifically, the process of acquiring the pressure distribution time-series matrix of the aircraft cockpit floor and seat contact surface, and extracting the three-dimensional coordinates of the human transient pressure center by performing spatiotemporal gradient analysis on the pressure distribution time-series matrix is ​​as follows: The discrete force signal sequence of the flexible piezoresistive array built into the cockpit floor and seat support surface is obtained; spatial surface interpolation is performed on the discrete force signal sequence to generate a continuous pressure topology surface; multi-channel clock synchronous sampling is performed on the continuous pressure topology surface to construct the pressure distribution time-series matrix; second-order spatial gradient feature extraction is performed along the three axes of the cockpit space in the pressure distribution time-series matrix to locate local pressure extremum clusters; and spatial torque balance equivalent transformation is performed on the local pressure extremum clusters to output the three-dimensional coordinates of the human transient pressure center.

[0030] In this implementation scheme, basic force information is first collected by sensor nodes deployed on the multi-dimensional contact surfaces of the cockpit. A surface interpolation algorithm is then used to smoothly transition the scattered point data into a continuous physical force surface covering the entire contact area. Subsequently, the system synchronously captures pressure slices from each channel at a fixed sampling period, constructing a pressure distribution time-series matrix that includes time-dimensional flow characteristics. To accurately pinpoint the core physical areas where the human body primarily bears weight, the system performs a second-order spatial gradient calculation on the pressure distribution time-series matrix in a three-dimensional coordinate system, identifying concentrated force areas with zero gradients and negative second derivatives, defining these as local pressure extremum clusters. Based on this, the system utilizes the principle of spatial moment balance, treating the human body as a three-dimensional rigid body model, and performs an equivalent mechanical transformation on the local pressure extremum clusters to obtain the three-dimensional coordinates of the transient pressure center of the human body that represents the overall physical center of mass shift trend of the occupant. The specific calculation formula is as follows: In the formula, : Represents the three-dimensional coordinate space vector of the human body's transient pressure center calculated at the current sampling moment; : Represents the sequence index parameter of the effective sensing nodes within the local pressure extreme value cluster; : Represents the total number of effective sensing nodes within the local pressure extreme value cluster; : indicates the first The absolute pressure scalar value collected by each effective sensing node at the current moment; : indicates the first The three-dimensional physical position reference vector of each effective sensing node in the cockpit fixed spatial coordinate system; : Represents the weighting coefficient of the corresponding node when performing spatial moment balance calculation. To accurately reflect the dominant physical effect of the main load-bearing components on the overall center of gravity shift, the weighting coefficient is dynamically assigned based on the local pressure percentage of each node, and its calculation formula is as follows: ,in and These represent the average pressure scalar and peak pressure scalar of the local pressure extreme cluster at the current moment, respectively. Through the above mechanical equivalent transformation process, the system can accurately reduce the complex surface contact force state of the human body over a large area to a single transient spatial coordinate point, providing a solid and accurate physical and mechanical benchmark for subsequent intention recognition.

[0031] Specifically, the process of performing frequency domain decoupling and separation of the low-frequency intention driving vector and the accompanying high-frequency jitter feature packet from the three-dimensional coordinates of the human transient pressure center is as follows: The three-dimensional coordinates of the human transient pressure center are arranged and reconstructed along the time axis to generate a motion trajectory sequence. Wavelet packet decomposition is performed on the motion trajectory sequence to extract multi-scale time-frequency domain feature coefficients. Based on the upper limit of the active force exertion frequency of human biomechanics, a cutoff frequency band splitting node is established. Low-frequency approximation coefficients below the cutoff frequency band splitting node are extracted and spatial vector reconstruction is performed to output the low-frequency intention driving vector. High-frequency detail coefficients above the cutoff frequency band splitting node are extracted and envelope detection convergence is performed to generate the accompanying high-frequency jitter feature packet.

[0032] In this implementation scheme, the previously obtained discrete spatial physical coordinates are smoothly connected in chronological order to form a continuous motion trajectory sequence that records the physical trend of human body displacement changes. Considering that the occupant's actions in the air environment not only involve active torso deflection and piloting intentions, but also inevitably superimposed with physiological muscle tremors caused by turbulence or psychological tension, the system introduces wavelet packet decomposition technology to perform full-band fine segmentation of the motion trajectory sequence, obtaining multi-scale time-frequency domain feature coefficients that combine temporal positioning and frequency resolution. To accurately remove unconscious noise at the physical level, the system introduces prior biomechanical features of the human body, setting the physical frequency threshold of the subjective torso force exertion as the cutoff frequency band splitting node. The system extracts the low-frequency approximation coefficients located below this splitting node, treating them as the occupant's true physical piloting intentions, and reconstructs them into a low-frequency intention driving vector using an inverse discrete wavelet reconstruction algorithm. The specific inverse spatial vector reconstruction formula is as follows: In the formula, : Represents the continuous low-frequency intention driving vector output after reconstruction; : Represents a variable that evolves over continuous time; : Represents the translation scale index variable for low-frequency approximation coefficients; : Indicates the total length of the approximation coefficients within the low-frequency band; : indicates the first The specific characteristic values ​​of each low-frequency approximation coefficient; : Represents the fundamental physical function used in the reconstruction process; : Represents the intrinsic translation step size parameter of the fundamental scale physics function; : Represents the physical scaling factor corresponding to the low-frequency band. Meanwhile, for high-frequency detail coefficients above the cutoff band split nodes, the system classifies them as involuntary human stress-induced physical tremors, and performs processing including detection to extract the energy fluctuation characteristics of the tremors. The specific envelope calculation formula is as follows: In the formula, : Represents the instantaneous physical energy envelope value within the associated high-frequency jitter characteristic packet output after detection and convergence; : Represents a high-frequency oscillation physical time series reconstructed from high-frequency detail coefficients; : Represents the Hilbert orthogonal integral transform operator performed on the time series. Through the above frequency domain decoupling and physical feature separation mechanism, the system completely eliminates the malicious physical interference of irrational physiological tremors of the human body on the attitude control of the aircraft at the underlying logic level, ensuring that the subsequent power distribution stage only responds to safe and smooth real piloting intentions.

[0033] Please see Figure 3Specifically, the process of mapping the low-frequency intention-driven vector to the local flow field coordinate system constructed by fusing the 3D terrain point cloud and the airflow particle motion vector collected by the airborne sensing device is as follows: Voxelization spatial downsampling is performed on the 3D terrain point cloud to establish a static terrain boundary matrix; the airflow particle motion vector is projected onto the non-occupied voxels of the static terrain boundary matrix to assign dynamic wind field attributes, thus constructing a local flow field coordinate system; the real-time pose angle parameters of the aircraft relative to the local flow field coordinate system are extracted to construct a spatial attitude transformation matrix; based on the spatial attitude transformation matrix, coordinate system alignment analysis is performed on the low-frequency intention-driven vector under the local reference of the cockpit to complete the spatial projection mapping to the local flow field coordinate system.

[0034] In this implementation scheme, to examine the localized piloting intentions generated by the occupants inside the cockpit within a real external three-dimensional physical space, the system first performs voxel-based spatial downsampling processing on the high-density three-dimensional terrain point cloud scanned by the airborne sensing equipment. This process divides the continuous physical space into discrete three-dimensional grids and assigns entity collision boundary attributes to grids containing obstacles, thereby establishing a static terrain boundary matrix. Subsequently, the system projects the airflow particle motion vectors acquired by external sensors onto the blank grids not occupied by obstacles in the static terrain boundary matrix, assigning dynamic wind field attributes to these non-occupied voxels, thus constructing a local flow field coordinate system that incorporates both terrain and airflow physical environment features. Based on this, the system extracts the roll, pitch, and yaw attitude parameters of the aircraft relative to the local flow field coordinate system, constructs a three-dimensional spatial attitude transformation matrix, and transforms the low-frequency intention driving vectors, originally based on the cockpit as a local reference frame, into the external physical space, completing coordinate system alignment resolution and spatial projection mapping. The specific spatial projection mapping calculation formula is as follows: In the formula, : indicates the first The low-frequency intentional target vector is mapped to the local flow field coordinate system by spatial projection at a discrete time step. : Represents the three-dimensional spatial attitude transformation matrix constructed based on real-time pose angle parameters; : Represents the roll attitude angle parameter of the aircraft in real time; : Represents the pitch attitude angle parameter of the aircraft in real time; : Represents the yaw attitude angle parameter of the aircraft's real-time position and attitude; : Represents the low-frequency intention driving vector extracted by the preceding sequence separation under the same discrete time step size; : Represents the spatial scale scaling factor during the coordinate system alignment and resolution process. The spatial scale scaling factor is determined by dynamically assigning a value based on the attenuation ratio between the aircraft's current altitude scalar and the standard sea-level atmospheric pressure reference value, in order to physically correct for the spatial deformation of the flow field coordinates at different flight altitudes. Through the above coordinate transformation and alignment resolution, the system successfully anchored the force direction of the human torso to the actual external wind field and terrain coordinate system, laying a unified spatial measurement benchmark for subsequent environmental safety assessments.

[0035] Specifically, the process of performing aerodynamic interference impedance analysis in the local flow field coordinate system by substituting the body dynamics envelope boundary into the aerodynamic impedance compensation matrix containing spatial disturbance rejection margin is as follows: Extending along the direction of the low-frequency intentional drive vector after spatial projection mapping, extract the directional airflow distribution gradient and terrain approximation distance in the local flow field coordinate system; performing joint boundary extremum determination on the directional airflow distribution gradient, terrain approximation distance, and the maximum lift coefficient and ultimate tilt angle in the body dynamics envelope boundary to extract the boundary risk factor; substituting the boundary risk factor into the preset aerodynamic impedance potential field model to deduce the normal repulsive force feedback tensor against the directional airflow distribution gradient and terrain approximation distance; performing dynamic vector synthesis on the normal repulsive force feedback tensor and the low-frequency intentional drive vector after spatial projection mapping to output the aerodynamic impedance compensation matrix containing spatial disturbance rejection margin.

[0036] In this implementation scheme, after completing the spatial projection mapping of the intent vector, to prevent the aircraft from blindly flying into the stall zone or colliding with obstacles due to the crew's complete compliance with the intention, the system will perform forward physical space extension detection along the mapped low-frequency intent drive vector direction within the local flow field coordinate system. Through this detection operation, the system extracts the directional airflow distribution gradient on the future flight path and the physical approximation distance with the surrounding terrain, thus sensing the wind shear intensity and physical collision threat ahead in advance. Subsequently, the system performs joint boundary extremum determination on the extracted environmental features and the maximum lift coefficient and ultimate roll angle defined within the dynamic envelope boundary of the aircraft itself, calculating a boundary risk factor that quantifies the danger level of the current operational intent. Next, the system substitutes the boundary risk factor into a preset aerodynamic impedance potential field model to simulate and generate a normal repulsive force feedback tensor that can physically push the aircraft away from the dangerous flow field region. Finally, the system performs dynamic vector synthesis of this normal repulsive force feedback tensor and the original spatial projection mapping intent vector, outputting an aerodynamic impedance compensation matrix that includes spatial disturbance rejection margin. The core calculation formula for the derived normal repulsive force feedback tensor is as follows: In the formula, : Represents the normal repulsive force feedback tensor generated by the deduction calculation; : Indicates the out-of-bounds risk factor extracted through joint boundary extreme value determination; : Represents the inherent repulsive potential energy gain coefficient of the preset aerodynamic impedance potential field model; : Represents the physical distance parameter for terrain approximation extracted along the direction of the intent vector; : Represents the system's preset physical safety distance parameter for hazard avoidance; : Represents the spatial vector of the directional airflow distribution gradient extracted along the forward detection direction. The boundary risk factor is established based on the ratio of the magnitude of the directional airflow distribution gradient vector to the maximum lift coefficient, combined with the reciprocal of the cosine of the limiting tilt angle, for a weighted calculation to accurately reflect the physical probability of the aircraft exceeding the aerodynamic envelope boundary. The preset physical avoidance safety distance parameter is determined by multiplying the aircraft's current absolute spatial flight velocity scalar by the time constant of the underlying rotor motor's electrical modulation response, ensuring that the impedance potential field model can pre-excite the repulsion tensor under high-speed flight conditions. By introducing aerodynamic impedance analysis and a repulsion tensor synthesis mechanism, the system not only ensures absolute flight trajectory safety in complex airflow environments at the physical level but also achieves flexible automatic wind deflection resistance, disturbance rejection, and collision avoidance correction without depriving the crew of control.

[0037] Specifically, the process of extracting the time-series amplitude sequence within the associated high-frequency jitter feature package and performing overlap feature extraction and matching between the time-series amplitude sequence and the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to obtain the human-machine coupling resonance band is as follows: The instantaneous energy envelope of each frequency band within the associated high-frequency jitter feature package is extracted to construct the time-series amplitude sequence. A dynamic state-space transformation is performed on the aerodynamic impedance compensation matrix to extract the expected body angle response frequency under the current flow field load. Cross-power spectral density estimation is performed on the time-series amplitude sequence and the expected body angle response frequency to generate a frequency domain coherent feature spectrum. The extreme value interval of the coherence coefficient is traced in the frequency domain coherent feature spectrum to define the overlapping frequency band. Dynamic closure constraints are applied to the upper and lower frequency boundaries of the overlapping frequency band to output the human-machine coupling resonance band.

[0038] In this implementation scheme, to detect the potential resonance risk between occupant physiological tremors and aircraft swaying, the energy change trend within the high-frequency tremor feature packet is first extracted to construct a time-series amplitude sequence. Simultaneously, the system inputs the previously acquired aerodynamic impedance compensation matrix into the aircraft's inherent dynamics model, and through state-space transformation, deduces the expected angular response frequency of the aircraft itself under the current complex flow field load. To quantify the overlap between these two independent oscillation sources in the frequency domain, the system introduces a cross-power spectral density estimation algorithm to generate a frequency-domain coherent feature spectrum. This process aims to reveal the physical coupling probability of the transfer of energy from unconscious human tremors to aircraft swaying; the core formula for calculating the coherent feature spectrum distribution is: In the formula, : Represents the coherence coefficient between human body shaking and bodily response at a specific analysis frequency; : Represents the independent frequency variable parameter in the frequency domain analysis process; : Represents the complex vector of cross-power spectral density between the time-series amplitude sequence and the expected body angular response frequency; : Represents the auto-success rate spectral density scalar of a time-series amplitude sequence; : Represents the self-success rate spectral density scalar of the expected aircraft angular response frequency. After obtaining the coherence coefficient across the entire frequency band, the system tracks the extreme value range of the coherence coefficient exceeding the preset coherence threshold in the frequency domain coherence characteristic spectrum, initially defining it as the overlapping frequency band. The preset coherence threshold is established based on the critical occurrence coherence values ​​of induced oscillation recorded in a large number of historical flight tests, plus a safety attenuation margin. Finally, the system performs dynamic closure constraints on the upper and lower frequency boundaries of the overlapping frequency band, that is, smoothly splices all continuous physical frequency bands that meet the coherence threshold condition into an envelope, thereby accurately delineating the human-machine coupling resonance band that is highly likely to cause aircraft instability. Through this series of frequency domain matching mechanisms, the system can accurately locate the fatal resonance physical interval, providing a clear range of action for subsequent targeted damping suppression.

[0039] Specifically, the process of analyzing the peak intensity and duration period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonant band is as follows: within the human-machine coupling resonant band, the peak intensity of the extreme value feature is analyzed and quantized by resonant energy spectral density aggregation. The duration period is extracted by tracing the coherent feature residence time span of the human-machine coupling resonant band along the time domain dimension. Nonlinear gain state matching is performed within the input damping adaptive attenuation mapping surface of the peak intensity and duration period. The gain adjustment extreme value of the output of the damping adaptive attenuation mapping surface is extracted and determined as the dynamic anti-oscillation attenuation coefficient.

[0040] In this implementation scheme, after establishing the human-machine coupling resonance band, the system needs to further quantify the physical threat level of this resonance state to flight safety. The system first performs a convergent analytical integral operation on the resonance energy spectral density within the defined resonance band frequency domain to quantify the extreme characteristics of the resonance energy accumulation, thereby outputting the peak intensity reflecting the intensity of the resonance. The specific quantification integral formula for the peak intensity is as follows: In the formula, : Represents the peak intensity scalar of the resonant energy of the human-machine coupled resonant band in the analytical output; : Represents the lower frequency limit closed boundary parameter of the human-machine coupling resonant band; : This represents the upper frequency limit closed boundary parameter of the human-machine coupling resonance band. Simultaneously, the system continuously tracks the physical residence time span of this coherent feature along the time evolution dimension, extracting the duration period parameter of the resonance phenomenon. This duration period is used to determine whether the occupant is experiencing brief physiological stress or is in a state of sustained panic and trembling. Subsequently, the system jointly inputs the peak intensity representing the resonance intensity and the duration representing the resonance duration into a pre-constructed damped adaptive attenuation mapping surface for nonlinear gain state matching. This mapping surface simulates the human-machine physical confrontation process, and its core nonlinear gain matching model is: In the formula, : Represents the dynamic anti-oscillation attenuation coefficient extracted and finally determined from the mapping surface. The value of this coefficient is between zero and one. : Represents the periodic scalar value indicating the duration of the resonance phenomenon being tracked and extracted; : Represents the oscillation energy damping penalty weight parameter set by the system; : Represents the nonlinear surface shape shaping index parameter. The oscillation energy damping penalty weight parameter is established based on a joint calibration of the aircraft's maximum controllable moment of inertia and the physical time of the underlying rotor response delay, ensuring that the applied damping force matches the aircraft's physical resistance capability. Through the above quantitative analysis and nonlinear mapping, the system can generate an adaptive attenuation coefficient in real time to suppress the physical response, based on the intensity of resonance induced by occupant stress. When the resonance is strong and persistent, this attenuation coefficient decreases rapidly, forcibly desensitizing the thrust response sensitivity of the underlying flight control system, thereby safely and smoothly blocking the deterioration path of vicious human-machine coupling oscillations and ensuring the absolute stability of the aircraft's overall attitude.

[0041] Please see Figure 4 Specifically, the process of inputting the low-frequency intention driving vector, aerodynamic impedance compensation matrix, and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint state-space optimization and generate the multirotor thrust allocation sequence within the control cycle is as follows: The low-frequency intention driving vector is set as the target tracking reference state of the nonlinear control solver; the aerodynamic impedance compensation matrix is ​​transformed into a state constraint boundary and fused into the rolling optimization objective function of the nonlinear control solver; the dynamic anti-oscillation attenuation coefficient is reconstructed into a dynamic penalty weight matrix of the nonlinear control solver at the control increment level; within the rolling optimization objective function, multivariable dynamic state trajectory rolling deduction optimization of the finite-time domain control variable sequence is performed based on the rigid body dynamics model; the optimal control input matrix after the deduction optimization convergence is extracted, and the dynamic allocation solution is performed to output the multirotor thrust allocation sequence within the control cycle.

[0042] In this implementation scheme, to transform the integrated state, which combines pilot intent and environmental constraints, into actual rotational speed commands for the underlying physical actuators, the system introduces a nonlinear control solver capable of handling multivariable physical constraints. The target tracking reference state represents the spatial physical trajectory that the aircraft should closely follow under ideal, undisturbed conditions; the rolling optimization objective function is a comprehensive mathematical cost model that evaluates and minimizes the physical deviation between the predicted physical behavior of the aircraft and the desired trajectory within a finite forward-sliding time window; and the state constraint boundary is the physical limit imposed by the flow field resistance and collision red lines that the aircraft cannot cross. The system first sets the low-frequency intent drive vector, representing the crew's true intent, as the target tracking reference state for the nonlinear control solver, using it as the dominant physical traction force guiding the aircraft's flight. Simultaneously, the system transforms the aerodynamic impedance compensation matrix obtained from the preceding analysis into the state constraint boundary and integrates it into the rolling optimization objective function, ensuring that any action that touches dangerous airflow or terrain boundaries will be rejected when predicting the future trajectory. Most importantly, the system reconstructs the previously extracted dynamic anti-oscillation attenuation coefficient into a dynamic penalty weight matrix for the nonlinear control solver at the control increment level. Based on this, the system performs multivariable dynamic state trajectory rolling deduction and optimization based on a rigid body dynamics model within the rolling optimization objective function, using a finite-time domain control variable sequence. The core rolling optimization objective function for this deduction and optimization is calculated as follows: In the formula, : Represents the cost scalar of the rolling optimization objective function constructed within the current control cycle; : Represents the total number of time steps for predicting the physical state within a finite time domain; : Represents the discrete sequence index parameter in the prediction time domain; : indicates the first generation generated based on the rigid body dynamics model. Multivariable dynamics prediction state matrix of the machine body; : Represents the first term derived from the low-frequency intention-driven vector. Step target tracking reference state matrix; : Represents the diagonal matrix of penalty weights for the system's state trajectory tracking error; : This parameter represents the total number of time steps that control the input increment sequence. : Represents the discrete sequence index parameter in the control time domain; : indicates the first The control input physical increment matrix for the step-by-step performance; : Represents the dynamic penalty weight matrix reconstructed from the dynamic anti-oscillation attenuation coefficient; : Represents the physical function of the penalty barrier constructed by introducing the aerodynamic impedance compensation matrix; : Represents the set of state constraint boundary spaces transformed from the aerodynamic impedance compensation matrix. The state trajectory tracking error penalty weight diagonal matrix is ​​established based on the physical mass distribution characteristics of the aircraft and the physical limit of the three-axis limiting angular velocity, using normalized joint diagonal assignment to ensure that the controller grants priority physical tracking authority to the core attitude response channel. The dynamic penalty weight matrix is ​​established based on the reciprocal of the preceding input dynamic anti-oscillation attenuation coefficient through adaptive scaling mapping. The ingenious function of this mechanism is that when the occupants experience significant high-frequency jitter due to panic, causing a sharp drop in the anti-oscillation attenuation coefficient, the norm of the dynamic penalty weight matrix increases non-linearly, causing the solver to impose extremely high physical penalties on any drastic changes in control increments when evaluating physical costs. This forces the solver to proactively abandon hasty correction actions and forcibly suppress abrupt changes in control commands during the physical solution phase. Ultimately, the system obtains the convergence trajectory that minimizes the cost scalar through online solving, extracts the optimal control input matrix within the first control step, and substitutes it into the rotor aerodynamic allocation configuration matrix of the aircraft to perform dynamic allocation calculation. The unified three-axis control torque physical quantity is decoupled and mapped into a multi-rotor thrust allocation sequence within the control cycle, thus completely completing the closed loop of underlying physical control from the physical center of gravity deflection of the crew to the stable flight of the aircraft.

[0043] Example 2; please refer to Figure 5 This system, used to execute a center-of-gravity driven precision attitude control system for an aircraft as described in the embodiments, includes: an intent decoupling module, used to acquire the pressure distribution time-series matrix of the aircraft cockpit floor and seat contact surface, perform spatiotemporal gradient analysis on the pressure distribution time-series matrix to extract the three-dimensional coordinates of the transient pressure center of the human body, and perform frequency domain decoupling on the three-dimensional coordinates of the transient pressure center of the human body to separate the low-frequency intent driving vector and the accompanying high-frequency jitter feature packet; and an aerodynamic anti-interference module, used to map the low-frequency intent driving vector to a local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and airflow particle motion vectors acquired by the airborne sensing equipment, and perform aerodynamic interference impedance analysis in the local flow field coordinate system by substituting the body dynamics envelope boundary. The system generates an aerodynamic impedance compensation matrix that includes spatial disturbance rejection margin; an oscillation suppression module is used to extract the time-series amplitude sequence within the associated high-frequency jitter feature package, and performs overlapping feature extraction and matching of the time-series amplitude sequence with the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to identify the human-machine coupling resonance band, and analyzes the peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band; a thrust allocation module is used to input the low-frequency intentional drive vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint optimization in the state space, generate a multi-rotor thrust allocation sequence within the control cycle, and convert the multi-rotor thrust allocation sequence into a pulse width modulation signal and send it to the underlying power mechanism.

[0044] In this implementation scheme, the intent decoupling module primarily undertakes the responsibility of perception and signal purification at the underlying level of human-machine interaction. This module directly faces the physical contact surface with the occupants, and its core function is to accurately extract the subjective force direction representing the pilot's true intention from the complex whole-body force state, while filtering out physiological tremors caused by fear of altitude or external turbulence as interference. Through this system design combining hardware acquisition and software decoupling, the aircraft can prevent ineffective or even dangerous physical actions from misleading the flight control host at the source, ensuring that the control commands received by the system are always stable and accurate piloting intentions, thus laying a reliable data input foundation for simplified operation. The aerodynamic disturbance rejection module primarily serves as a physical boundary safety guardian for the aircraft's space navigation. This module does not blindly execute the occupants' gravity deflection commands, but instead projects the occupants' piloting intentions into a realistic three-dimensional environment including terrain obstacles and airflow motion for safety rehearsal. Its core function is to assess in real time whether the current center-of-gravity drive command will cause a collision or stall hazard, and automatically generate a space impedance margin to correct the original command when a potential environmental threat is detected. This modular design enables the aircraft to maintain its fundamental defensive capabilities against sudden crosswinds and actively avoid physical collisions, even in a control mode entirely reliant on crew center of gravity displacement. The oscillation suppression module is primarily responsible for breaking the vicious cycle of attitude loss caused by human-machine coupling. This module constantly monitors the crew's high-frequency involuntary shaking and compares it in real-time with the aircraft's own swaying state. Its core function is to accurately capture the dangerous critical point between crew tension and trembling and aircraft resonance, and dynamically output damping attenuation strategies based on the severity of the resonance. Through the physical intervention of this module, the system can decisively blunt the underlying dynamic response in the early stages of human-machine resonance, forcibly calming the violent shaking of the aircraft, greatly improving the control tolerance and overall stability of the aircraft for ordinary pilots facing sudden high-altitude situations. The thrust distribution module is the core decision-making and physical execution hub of the entire center of gravity-driven attitude control system. This module globally coordinates the pure piloting intention, environmental safety compensation, and anti-oscillation damping parameters, calculating the optimal flight attitude control strategy online while meeting all dynamic constraints and physical red lines. Its core function is to solve the problem of underlying power allocation under multi-variable conflict, seamlessly transforming the abstract comprehensive spatial state into physical speed control signals specific to each rotor motor, and finally completing the hardware execution closed loop from human instinctive attitude deflection to precise displacement of the aircraft in the three-dimensional low-altitude flow field.

[0045] In summary, this application has at least the following effects: A method and system for precise attitude control of aircraft driven by the center of gravity is proposed. By collecting cockpit pressure distribution data and performing frequency domain decoupling, the system accurately separates the occupant's true low-frequency piloting intentions from unconscious high-frequency physiological vibrations. The true intentions are then mapped onto a local flow field coordinate system incorporating terrain and airflow for aerodynamic interference impedance deduction. An adaptive anti-oscillation attenuation coefficient is extracted by combining the frequency domain coherence characteristics of high-frequency vibrations and the expected aircraft response. Finally, the aforementioned multi-dimensional physical constraints are jointly input into a nonlinear control solver to perform dynamic state-space optimization and underlying thrust allocation. This scheme completely overcomes the physical defects of traditional mechanical control, such as high learning barriers and susceptibility to human stress-induced tremors in purely haptic control. Based on establishing a direct mechanical mapping between the dynamic shift of the occupant's physical center of gravity and the aircraft's spatial attitude, it not only actively avoids aerodynamic stall and collision risks in complex low-altitude flow fields at the physical level but also forcibly cuts off the vicious path of human-induced resonance instability from the dimension of biomechanical and aerodynamic coupling. This achieves ultimate smoothness and absolute physical safety for aircraft navigation in complex three-dimensional space under a zero-threshold, minimalist piloting mode.

[0046] 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 precise attitude control of a center-of-gravity driven aircraft, characterized in that, Includes the following steps: S1. Collect the pressure distribution time series matrix of the aircraft cockpit floor and seat contact surface, perform spatiotemporal gradient analysis on the pressure distribution time series matrix to extract the three-dimensional coordinates of the human transient pressure center, and perform frequency domain decoupling on the three-dimensional coordinates of the human transient pressure center to separate the low-frequency intention driving vector and the accompanying high-frequency jitter feature package. S2. Map the low-frequency intention driving vector to the local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and the airflow particle motion vector collected by the airborne sensing device. Substitute the body dynamics envelope boundary into the local flow field coordinate system to perform aerodynamic interference impedance analysis and generate an aerodynamic impedance compensation matrix containing spatial disturbance rejection margin. S3. Extract the time-series amplitude sequence from the associated high-frequency jitter feature package, perform overlapping feature extraction and matching of the time-series amplitude sequence and the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to find the human-machine coupling resonance band, and analyze the peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band. S4. Input the low-frequency intention driving vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint optimization in the state space, generate the multi-rotor thrust allocation sequence within the control cycle, and convert the multi-rotor thrust allocation sequence into a pulse width modulation signal and send it to the underlying power mechanism.

2. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 1, characterized in that: The specific process of collecting the pressure distribution time-series matrix of the contact surface between the aircraft cockpit floor and the seat, and performing spatiotemporal gradient analysis on the pressure distribution time-series matrix to extract the three-dimensional coordinates of the human transient pressure center is as follows: The discrete force signal sequence of the flexible piezoresistive array built into the cockpit floor and seat support surface is obtained, and the discrete force signal sequence is approximated by spatial surface interpolation to generate a continuous pressure topology surface. Multi-channel clock synchronization sampling is performed on the continuous pressure topology to construct a pressure distribution time series matrix. Second-order spatial gradient feature extraction is performed along the three axes of the cabin space in the pressure distribution time series matrix to locate local pressure extremum clusters. Perform a spatial torque balance equivalent transformation on the local pressure extreme value cluster to output the three-dimensional coordinates of the human body transient pressure center.

3. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 1, characterized in that: The specific process of performing frequency domain decoupling and separation of the low-frequency intention driving vector and the accompanying high-frequency jitter feature package from the three-dimensional coordinates of the human body's transient pressure center is as follows: The three-dimensional coordinates of the human body's transient pressure center are arranged and reconstructed along the time axis to generate a motion trajectory sequence. Wavelet packet decomposition is then performed on the motion trajectory sequence to extract multi-scale time-frequency domain feature coefficients. Based on the upper limit of the active force exertion frequency of human biomechanics, cutoff frequency band split nodes are established. Low-frequency approximation coefficients below the cutoff frequency band split nodes are extracted and spatial vector reconstruction is performed to output low-frequency intention driving vectors. High-frequency detail coefficients above the cutoff frequency band split nodes are extracted and envelope detection convergence is performed to generate associated high-frequency jitter feature packets.

4. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 1, characterized in that: The specific process of mapping the low-frequency intention-driven vector to the local flow field coordinate system constructed by fusing the 3D terrain point cloud and airflow particle motion vectors collected by the airborne sensing device is as follows: Voxelization spatial downsampling is performed on the 3D terrain point cloud to establish a static terrain boundary matrix. The airflow particle motion vector is projected onto the non-occupied voxels of the static terrain boundary matrix to give dynamic wind field attributes and construct a local flow field coordinate system. Extract the real-time pose angle parameters of the aircraft relative to the local flow field coordinate system, construct a spatial attitude transformation matrix, and perform coordinate system alignment analysis on the low-frequency intention drive vector under the local reference of the cockpit based on the spatial attitude transformation matrix to complete the spatial projection mapping to the local flow field coordinate system.

5. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 4, characterized in that: The specific process of performing aerodynamic interference impedance analysis in the local flow field coordinate system by substituting the body dynamics envelope boundary and generating the aerodynamic impedance compensation matrix including spatial disturbance rejection margin is as follows: The heading airflow distribution gradient and terrain approximation distance within the local flow field coordinate system are extracted by extending along the direction of the low-frequency intentional driving vector after spatial projection mapping. The maximum lift coefficient and the ultimate tilt angle in the heading airflow distribution gradient, terrain approach distance and the airframe dynamics envelope boundary are used to perform joint boundary extreme value determination to extract the boundary risk factor; Substitute the risk factor of crossing the boundary into the preset aerodynamic impedance potential field model to deduce the normal repulsive force feedback tensor that counteracts the gradient of the heading airflow distribution and the approach distance to the terrain. The low-frequency intentional driving vector, mapped from the normal repulsive force feedback tensor to the spatial projection, is used to perform dynamic vector synthesis to output an aerodynamic impedance compensation matrix that includes spatial disturbance rejection margin.

6. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 1, characterized in that: The specific process of extracting the time-series amplitude sequence from the associated high-frequency jitter feature packet and performing overlap feature extraction and matching between the time-series amplitude sequence and the expected body angular response frequency mapped by the aerodynamic impedance compensation matrix to determine the human-machine coupling resonance band is as follows: The instantaneous energy envelopes of each frequency band within the associated high-frequency jitter feature package are extracted to construct a time-series amplitude sequence. A dynamic state-space transformation is performed on the aerodynamic impedance compensation matrix to extract the expected body angular response frequency under the current flow field load. Cross-power spectral density estimation is performed on the time-series amplitude sequence and the expected body angular response frequency to generate a frequency domain coherent characteristic spectrum. The extreme value interval of the coherence coefficient is traced in the frequency domain coherent characteristic spectrum to define the overlapping frequency band. Dynamic closure constraints are applied to the upper and lower limit frequency boundaries of the overlapping frequency band to output the human-machine coupling resonance band.

7. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 6, characterized in that: The specific process of analyzing the peak intensity of the human-machine coupling resonance band and the dynamic anti-oscillation attenuation coefficient of the continuous period output is as follows: Within the human-machine coupling resonant band, the peak intensity of the extreme value feature is output by performing resonant energy spectral density aggregation analytical quantization, and the duration period is extracted by tracing the coherent feature residence time span of the human-machine coupling resonant band along the time domain dimension. The peak intensity is matched with the nonlinear gain state within the continuous period input damping adaptive attenuation mapping surface, and the gain adjustment extreme value of the output of the damping adaptive attenuation mapping surface is extracted and determined as the dynamic anti-oscillation attenuation coefficient.

8. The method for precise attitude control of a center-of-gravity driven aircraft according to claim 1, characterized in that: The specific process of inputting the low-frequency intentional drive vector, aerodynamic impedance compensation matrix, and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint optimization in the state space and generate the multi-rotor thrust allocation sequence within the control cycle is as follows: The low-frequency intention driving vector is set as the target tracking reference state of the nonlinear control solver, and the aerodynamic impedance compensation matrix is ​​transformed into the state constraint boundary and fused into the rolling optimization objective function of the nonlinear control solver. The dynamic anti-oscillation attenuation coefficient is reconstructed into a dynamic penalty weight matrix at the control increment level of the nonlinear control solver. Within the rolling optimization objective function, the multivariable dynamic state trajectory of the finite-time domain control variable sequence is rolled and optimized based on the rigid body dynamics model. Extract the optimal control input matrix after the deduction and optimization convergence, perform dynamic allocation calculation, and output the multi-rotor thrust allocation sequence within the control cycle.

9. A center-of-gravity driven precision attitude control system for an aircraft, used to execute the center-of-gravity driven precision attitude control method for an aircraft as described in any one of claims 1-8, characterized in that, include: The intent decoupling module is used to collect the pressure distribution time series matrix of the aircraft cockpit floor and seat contact surface, perform spatiotemporal gradient analysis on the pressure distribution time series matrix to extract the three-dimensional coordinates of the human transient pressure center, and perform frequency domain decoupling on the three-dimensional coordinates of the human transient pressure center to separate the low-frequency intent driving vector and the accompanying high-frequency jitter feature package. The aerodynamic anti-interference module is used to map the low-frequency intention drive vector to the local flow field coordinate system constructed by fusing the three-dimensional terrain point cloud and the airflow particle motion vector collected by the airborne sensing equipment. It then substitutes the body dynamics envelope boundary into the local flow field coordinate system to perform aerodynamic interference impedance analysis and generate an aerodynamic impedance compensation matrix that includes spatial anti-interference margin. The oscillation suppression module is used to extract the time-series amplitude sequence within the associated high-frequency jitter feature package, perform overlapping feature extraction and matching of the time-series amplitude sequence with the expected body angle response frequency mapped by the aerodynamic impedance compensation matrix to identify the human-machine coupling resonance band, and analyze the peak intensity and continuous period output dynamic anti-oscillation attenuation coefficient of the human-machine coupling resonance band. The thrust allocation module is used to input the low-frequency intention drive vector, aerodynamic impedance compensation matrix and dynamic anti-oscillation attenuation coefficient into the nonlinear control solver to perform joint optimization in the state space, generate the multi-rotor thrust allocation sequence within the control cycle, and convert the multi-rotor thrust allocation sequence into a pulse width modulation signal and send it to the underlying power mechanism.

Images