Unmanned aerial vehicle hoisting load attitude solution system based on force distribution of whole circumference of hoisting point

Abstract

The present application relates to unmanned aerial vehicle hoisting and intelligent hoisting control technical field, especially in based on the whole stress distribution of the hoisting point unmanned aerial vehicle hoisting load attitude solution system. The annular bearing base of the system connects the hoisting equipment and the hoisted load and transmits the load force; the whole stress sensing unit obtains the stress signal of different directions of the base to form the circumferential stress field; the signal preprocessing unit synchronously collects the signal, processes the consistency and eliminates the common mode noise, and extracts the dynamic characteristic vector; the vector reconstruction unit reconstructs the pose parameters of the resultant force vector according to the stress distribution law; the attitude solution unit obtains the swing angle, direction and angular velocity of the load according to the parameters such as eccentricity and equivalent hoisting length. The present application realizes the high-precision real-time monitoring of the spatial attitude of the hoisted load through the stress distribution reconstruction of the whole circumferential direction of the hoisting point, effectively suppresses the dynamic interference of the unmanned aerial vehicle itself, and provides reliable data support for the autonomous swing damping and stable control of the unmanned aerial vehicle hoisting operation.

Description

Technical Field

[0001] This invention relates to the field of aircraft attitude control technology, and in particular to a UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point. Background Technology

[0002] During drone-based hoisting and lifting operations, the load being hoisted and the flight platform constitute a typical flexible multibody dynamics system. The hoisted cargo is susceptible to the influence of external random airflow, drastic changes in the drone's attitude, and inertial forces caused by the aircraft's acceleration and deceleration, making it prone to multidimensional spatial swaying.

[0003] Currently, existing solutions for hoisting load attitude detection and swing control mainly suffer from the following technical shortcomings:

[0004] In existing technologies, some solutions rely on computer vision, global positioning systems (GPS), or inertial measurement units (IMUs) installed on the payload. Vision solutions are prone to target loss in environments with strong light, low light, smoke, dust, or obstruction by slings; GPS cannot obtain high-precision coordinates in shielded conditions such as tunnels, deep valleys, or tall building complexes; while payload-side IMUs that transmit data wirelessly face problems such as cumbersome pairing, limited battery life, and data drift caused by high-frequency vibration interference.

[0005] Traditional force measurement systems are mostly based on load cell designs, primarily focusing on monitoring the total weight of the load. However, load oscillation is a dynamic vector change process, and traditional single-axis or dual-axis force sensors can only acquire the magnitude of the force, unable to identify the deflection pattern of the force's line of action relative to the center of the suspension point. Due to the lack of coupled sensing of the off-center load direction, dynamic oscillation angle, and angular velocity, the back-end controller struggles to construct an accurate compensation model.

[0006] Conventional suspension-point force measurement structures often employ single-point suspension or symmetrically arranged discrete sensing points (such as four sensors distributed in a cross shape). In this layout, the force transmission path is confined to a limited number of geometric support points, resulting in a significant blind spot in force sensing across the circumferential space. Because it is impossible to achieve continuous 360° circumferential stress field sensing, the system struggles to detect subtle changes in eccentricity. When faced with irregular oscillations or complex motions, the accuracy and response frequency of spatial attitude calculation are far from meeting the stable control requirements of high-dynamic environments. Summary of the Invention

[0007] To overcome the above deficiencies, this invention provides a UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point. The aim is to achieve high-frequency, high-precision spatial attitude calculation using only the force signal distribution characteristics of the hoisting point itself, thereby improving the stability and environmental adaptability of the UAV hoisting system in all scenarios.

[0008] This invention provides the following technical solution: a UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point, comprising:

[0009] The ring-shaped load-bearing base is used to connect the hoisting equipment and the hoisted load and to transmit the load force;

[0010] A full-circumference force sensing unit is set in the circumference of the ring-shaped bearing base to acquire the original force signals of the suspension point structure in different circumferential directions, so as to form a circumferential stress field reflecting the load stress state.

[0011] The signal preprocessing unit is used to synchronously acquire and process the original force signals from all directions, and extract the dynamic feature vector of the circumferential stress field.

[0012] The vector reconstruction unit is used to reconstruct the pose parameters of the resultant force vector based on the amplitude distribution law of the feature vector in the circumferential direction, wherein the pose parameters include magnitude, direction and eccentricity.

[0013] The attitude calculation unit is used to calculate the pose parameters of the resultant force vector to obtain the spatial attitude data of the load relative to the hoisting equipment.

[0014] The communication output module is used to send the calculated spatial attitude data to the control equipment in real time.

[0015] Preferably, in the full-circumference force sensing unit, the step of forming a circumferential stress field reflecting the load stress state includes:

[0016] By using discrete sensing points arranged circumferentially along the ring-shaped bearing substrate, the original force signals from different directions are acquired synchronously.

[0017] Based on the structural geometric parameters and stiffness distribution characteristics of the ring-shaped bearing substrate, the original force signals in each direction are compensated and corrected to eliminate the strain response differences caused by the non-circular structure.

[0018] The corrected force signal is used as the sampled value, and a curve is fitted using a generalized periodic function containing higher-order harmonic components to reconstruct a continuous closed-loop circumferential stress distribution field in the spatial dimension.

[0019] Preferably, in the signal preprocessing unit, the steps of synchronously acquiring and uniformly processing the original force signals from all directions include:

[0020] In the unloaded state of hoisting, the static output values ​​of sensing points at different orientations are acquired synchronously as the zero-point reference of the original force signal;

[0021] By using the preset gain deviation coefficient between each sensing point, the original force signal acquired dynamically is corrected so that the amplitude changes of the signal in each direction tend to be consistent under the same force change.

[0022] Common-mode noise is eliminated for the original force signals in all directions, and the signals are strictly synchronized and aligned in the time dimension to eliminate the phase difference in the dynamic force process.

[0023] Preferably, the step of common-mode noise cancellation for the original force signals in each direction includes:

[0024] The original force signals from different directions are collected to form a multi-dimensional observation matrix;

[0025] Principal component analysis was performed on the multidimensional observation matrix, and the principal component with the largest eigenvalue was extracted as the global common mode component.

[0026] The signal after eliminating common-mode noise is obtained by subtracting the projection of the global common-mode component onto each channel from the original force signals in each direction.

[0027] Preferably, in the signal preprocessing unit, the step of extracting the dynamic feature vector of the circumferential stress field includes:

[0028] Calculate the rate of change of sensor signals in different orientations over time to obtain dynamic incremental data reflecting the load swing trend;

[0029] Arrange the original signals from each direction and the dynamic incremental data according to the circumferential angular position to form a spatial vector group containing the circumferential angle, force amplitude and amplitude change rate;

[0030] From the spatial vector set, common interference components in different directions are removed, and the dominant feature vector that characterizes the change in the eccentric intensity and direction of the circumferential stress field is extracted as the dynamic feature vector.

[0031] Preferably, the step of removing common interference components from the spatial vector set includes:

[0032] Obtain the spatial vector distribution of the load under a preset reference state as a dynamic background reference;

[0033] The spatial vector group is subjected to point-by-point vector difference with the dynamic background reference to suppress static features caused by environmental factors or structural weight.

[0034] Calculate the correlation coefficient between each signal and the mean of all signals, and remove global pulsation components whose correlation coefficient is greater than a preset threshold.

[0035] Preferably, in the vector reconstruction unit, the step of reconstructing the pose parameters of the resultant force vector includes:

[0036] Obtain the force amplitude sequence corresponding to the dynamic feature vector in different directions in the circumferential direction;

[0037] Curve fitting is performed on the force amplitude sequence to obtain the circumferential force distribution curve;

[0038] The direction of the peak value of the force distribution curve is taken as the direction of the resultant force vector, and the magnitude of the peak value of the force distribution curve is taken as the magnitude of the resultant force vector.

[0039] Based on the difference between the peak and valley values ​​of the force distribution curve, and combined with the calibration coefficient of the annular bearing substrate, the eccentricity of the resultant force vector is calculated.

[0040] Preferably, in the attitude calculation unit, the step of calculating the pose parameters of the resultant force vector includes:

[0041] The swing angle of the load is calculated based on the eccentricity and the effective lever arm radius of the annular bearing base.

[0042] The direction angle of the resultant force vector is directly used as the swing direction of the load;

[0043] Based on the rate of change of the direction angle of the resultant force vector with time, the oscillation angular velocity of the load is calculated, and the direction angle sequence is filtered.

[0044] The swing angle, swing direction, and swing angular velocity are output as spatial attitude data.

[0045] Preferably, in the communication output module, the communication output module sends data to the UAV flight controller via a CAN bus, RS485, or serial interface.

[0046] The present invention has the following beneficial effects:

[0047] 1. This invention does not rely on vision, GPS or load-end sensors. It can calculate the attitude only through the circumferential stress field of the suspension point itself, which fundamentally solves the impact of complex lighting, electromagnetic shielding and signal blockage on detection accuracy. It significantly improves the system's environmental adaptability and operational continuity in extreme working conditions such as tunnels and indoor environments.

[0048] 2. This invention, through a ring-shaped bearing base and a vector reconstruction algorithm, can transform discrete force signals into continuous circumferential distribution curves. It can not only accurately monitor the load weight, but also calculate high-dimensional dynamic parameters such as swing angle, direction and angular velocity in real time, providing more complete and refined data feedback for the active anti-sway control of UAVs.

[0049] 3. This invention utilizes principal component analysis to eliminate common-mode noise and combines it with geometric compensation to correct non-uniform strain, effectively isolating the aircraft from vibration and environmental interference. Through real-time quantification of eccentricity, it can sensitively monitor the off-center load state and structural fatigue risk of the hoisting system, significantly enhancing the safety early warning capability throughout the entire UAV hoisting operation process. Attached Figure Description

[0050] Figure 1 This is a structural diagram of the UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point proposed in this invention;

[0051] Figure 2 A simplified schematic diagram of an annular bearing base provided in an embodiment of the present invention includes: an upper plate (1) connected to a load-bearing hook, a load-bearing hook (2), an upper surface of the annular bearing base (3), a distributed sensor (4), and an upper surface of the annular bearing base (5).

[0052] Figure 3 This is a real-time flowchart of a UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point, provided in an embodiment of the present invention. Detailed Implementation

[0053] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Example 1

[0055] In a first embodiment of the present invention, the present invention provides a UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point, such as... Figure 1 As shown, it includes:

[0056] The ring-shaped load-bearing base is used to connect the hoisting equipment and the hoisted load and to transmit the load force;

[0057] Specifically, the annular bearing base is the main structure for hoisting and bearing loads. Distributed sensors mounted on it form a circumferential force distribution measurement area along the circumference of the annular bearing base, used to collect stress, strain, or pressure distribution data within a 360° range, forming a complete force distribution curve. The force distribution measurement area can include any one of the following: a continuous strain sensing area, a multi-segment partitioned sensing area, or a multi-point discrete sensing array. The number of sensing units, their spacing, and whether their arrangement is symmetrical or asymmetrical do not affect the acquisition of the circumferential force distribution and the reconstruction of the resultant force vector. The annular bearing base can include any one of the following: a circular lifting ring, a polygonal lifting ring, a flange-type lifting point, or a shackle-type structure. As long as it has circumferential force bearing capacity and can achieve circumferential force distribution measurement, it falls within the protection range. This embodiment provides an annular bearing base such as... Figure 2 As shown, it includes: an upper plate 1 connected to a load-bearing hook, a load-bearing hook 2, an upper surface of an annular load-bearing base 3, a distributed sensor 4, and an upper surface of an annular load-bearing base 5.

[0058] A full-circumference force sensing unit is set in the circumference of the ring-shaped bearing base to acquire the original force signals of the suspension point structure in different circumferential directions, so as to form a circumferential stress field reflecting the load stress state.

[0059] Preferably, in the full-circumference force sensing unit, the step of forming a circumferential stress field reflecting the load stress state includes:

[0060] By using discrete sensing points arranged circumferentially along the ring-shaped bearing substrate, the original force signals from different directions are acquired synchronously.

[0061] Based on the structural geometric parameters and stiffness distribution characteristics of the ring-shaped bearing substrate, the original force signals in each direction are compensated and corrected to eliminate the strain response differences caused by the non-circular structure.

[0062] The corrected force signal is used as the sampled value, and a curve is fitted using a generalized periodic function containing higher-order harmonic components to reconstruct a continuous closed-loop circumferential stress distribution field in the spatial dimension.

[0063] Specifically, the full-circumference force sensing unit acquires the force state through multiple sensing points arranged on the circumferential surface of the annular bearing substrate. The output of each sensing point is sampled synchronously through multiple channels to ensure the phase consistency of the original force signals in all directions on the time axis.

[0064] Because the annular load-bearing base may exhibit non-circular geometric shapes in practical applications, such as polygonal lifting rings or flange structures, the moment of inertia and stiffness characteristics of its cross-sections differ at different locations. This implementation system corrects the original force signal using a pre-stored geometric compensation matrix. The correction process is as follows: Let the first... The original force signal at each sensing point is The geometric correction factor corresponding to this site is The stiffness compensation amount is The corrected force signal Calculated using the following formula:

[0065] ;

[0066] in, and These are characteristic parameters obtained through pre-calibration of the annular bearing substrate, used to eliminate the influence of structural non-uniformity on signal amplitude.

[0067] After acquiring a set of discretely distributed corrected force signals, the system uses a generalized periodic function containing higher-order harmonic components for curve fitting, thereby reconstructing a spatially continuous circumferential stress distribution field. This fitting process employs a Fourier series expansion, and the circumferential stress distribution field function... Represented as:

[0068] ;

[0069] in, Represents the circumferential azimuth of the annular bearing substrate; The DC component represents the global average force level. The harmonic order is... and For the first Amplitude coefficient of the first harmonic; The fitting order is determined based on the number of sensing sites, typically set to less than half the total number of sensing sites to satisfy the sampling theorem. For a standard circular substrate, the fitting function is dominated by the first-order fundamental component. For polygonal or detachable structures, stress concentration characteristics caused by geometric abrupt changes will be reflected in higher-order harmonic components. The coefficients for each order are calculated using the least squares method. and So that each discrete sampling point With function value The sum of squared residuals is minimized. The final circumferential stress distribution field is a set of values ​​related to the angle. The continuous function describes the real-time distribution of load force in the entire circumference of the annular bearing substrate, providing a continuous spatial data foundation for subsequent vector reconstruction.

[0070] The signal preprocessing unit is used to synchronously acquire and process the original force signals from all directions, and extract the dynamic feature vector of the circumferential stress field.

[0071] Preferably, in the signal preprocessing unit, the steps of synchronously acquiring and uniformly processing the original force signals from all directions include:

[0072] In the unloaded state of hoisting, the static output values ​​of sensing points at different orientations are acquired synchronously as the zero-point reference of the original force signal;

[0073] By using the preset gain deviation coefficient between each sensing point, the original force signal acquired dynamically is corrected so that the amplitude changes of the signal in each direction tend to be consistent under the same force change.

[0074] Common-mode noise is eliminated for the original force signals in all directions, and the signals are strictly synchronized and aligned in the time dimension to eliminate the phase difference in the dynamic force process.

[0075] Preferably, the step of common-mode noise cancellation for the original force signals in each direction includes:

[0076] The original force signals from different directions are collected to form a multi-dimensional observation matrix;

[0077] Principal component analysis was performed on the multidimensional observation matrix, and the principal component with the largest eigenvalue was extracted as the global common mode component.

[0078] The signal after eliminating common-mode noise is obtained by subtracting the projection of the global common-mode component onto each channel from the original force signals in each direction.

[0079] Specifically, the signal preprocessing unit first establishes a consistency benchmark. Under no-load conditions before the lifting operation begins, the system initiates a self-calibration procedure to synchronously acquire... Static output value of each sensing point The static output value includes sensor mounting preload, inherent circuit offset, and zero-point drift, serving as the zero-point reference for each signal. During the dynamic acquisition phase, due to manufacturing tolerances in the sensitivity of different sensors, the system utilizes a pre-stored gain deviation coefficient. Amplitude correction is performed on the real-time signal. Let... Time of the first The raw force signal collected by the road sensor site is The signal after the initial correction Represented as:

[0080] ;

[0081] This step ensures that the sensing signals from all directions have a linear and consistent amplitude response under the same mechanical excitation.

[0082] To eliminate phase differences during dynamic stress processes, the acquired digital signal sequences are aligned using a time-domain algorithm to compensate for minor time offsets caused by signal transmission link delays, ensuring that data from all channels reflects the true physical field state at the same point in time. The specific implementation process for common-mode noise cancellation addresses the common-mode noise prevalent in UAV hoisting operations caused by high-frequency vibrations of the UAV rotor and vertical lift acceleration. The system employs principal component analysis to remove this noise. First, within a preset time window, the system... The original force signals from each direction are used to construct a multi-dimensional observation matrix. If the number of sampling points is Then the matrix The dimension is , means as follows:

[0083] ;

[0084] Subsequently, the system calculates the matrix. The covariance matrix is ​​calculated, and eigenvalues ​​and their corresponding eigenvectors are obtained through eigenvalue decomposition. In the drone hoisting scenario, due to the high directional consistency and temporal synchronization of the forces exerted by the overall motion of the drone on all sensing points, the principal component with the largest eigenvalue... This represents the global common-mode component. Finally, the system calculates the projection of this global common-mode component onto each sensing channel. Let the first... The projection factor of the channel is The effective signal after eliminating common-mode noise Obtained through subtraction:

[0085] ;

[0086] This process eliminates global force disturbances that are independent of the load swing direction, in order to retain the differential components with orientation differences caused by load eccentricity.

[0087] Preferably, in the signal preprocessing unit, the step of extracting the dynamic feature vector of the circumferential stress field includes:

[0088] Calculate the rate of change of sensor signals in different orientations over time to obtain dynamic incremental data reflecting the load swing trend;

[0089] Arrange the original signals from each direction and the dynamic incremental data according to the circumferential angular position to form a spatial vector group containing the circumferential angle, force amplitude and amplitude change rate;

[0090] From the spatial vector set, common interference components in different directions are removed, and the dominant feature vector that characterizes the change in the eccentric intensity and direction of the circumferential stress field is extracted as the dynamic feature vector.

[0091] Preferably, the step of removing common interference components from the spatial vector set includes:

[0092] Obtain the spatial vector distribution of the load under a preset reference state as a dynamic background reference;

[0093] The spatial vector group is subjected to point-by-point vector difference with the dynamic background reference to suppress static features caused by environmental factors or structural weight.

[0094] Calculate the correlation coefficient between each signal and the mean of all signals, and remove global pulsation components whose correlation coefficient is greater than a preset threshold.

[0095] Specifically, in the signal preprocessing unit, after obtaining the force signals in each direction after consistency processing, the system first calculates the rate of change of the sensing signals in each direction over time using first-order difference. Let... Time of the first The processed force signals at each direction are The sampling period is Then the rate of change of amplitude in that direction Represented as:

[0096] ;

[0097] Income This constitutes dynamic incremental data, used to reflect the instantaneous evolution trend of the stress field caused by load oscillation.

[0098] Subsequently, the original force amplitudes in each direction were determined. Dynamic incremental data With the corresponding installation azimuth angle Perform associations to construct spatial vector groups This vector set can be represented at any sampling time as a set containing... A set of vectors with n elements, each element containing three-dimensional features:

[0099] ;

[0100] This vector group transforms scattered electrical signals into a structured dataset with spatial orientation and temporal trend attributes.

[0101] To further purify the dynamic feature vectors, the system employs a two-step method to eliminate common interference components in the spatial vector set. The first step utilizes a dynamic background benchmark to suppress static features. The system pre-records the spatial vector distribution under vertical static equilibrium conditions, denoted as... During real-time processing, the current spatial vector group is... and Perform point-by-point vector difference:

[0102] ;

[0103] This differential operation eliminates the quasi-static stress background caused by the self-weight of the ring-shaped bearing base, the constant gravity of the load, and the ambient temperature, allowing the system to respond centrally to the variables generated by the oscillation.

[0104] The second step involves eliminating global pulsation components through correlation analysis. Global pulsation is typically generated by the overall jitter of the UAV during the hoisting process, manifesting as synchronous fluctuations in the time series of all azimuth signals. The system first calculates the mean sequence of all azimuth signals within the current sampling window. :

[0105] ;

[0106] Then, calculate each signal. With mean sequence correlation coefficient .for Greater than the preset threshold The components that are identified as global pulsation components are then removed. The threshold value is... The value range is typically set to 0.85 to 0.95. After removing common interference, the spatial vector group undergoes regularization to form the final dominant feature vector. This vector represents the eccentric intensity and direction of the circumferential stress field and eliminates spurious features caused by environmental and organismal motion, serving as input data for subsequent vector reconstruction units.

[0107] The vector reconstruction unit is used to reconstruct the pose parameters of the resultant force vector based on the amplitude distribution law of the feature vector in the circumferential direction, wherein the pose parameters include magnitude, direction and eccentricity.

[0108] Preferably, in the vector reconstruction unit, the step of reconstructing the pose parameters of the resultant force vector includes:

[0109] Obtain the force amplitude sequence corresponding to the dynamic feature vector in different directions in the circumferential direction;

[0110] Curve fitting is performed on the force amplitude sequence to obtain the circumferential force distribution curve;

[0111] The direction of the peak value of the force distribution curve is taken as the direction of the resultant force vector, and the magnitude of the peak value of the force distribution curve is taken as the magnitude of the resultant force vector.

[0112] Based on the difference between the peak and valley values ​​of the force distribution curve, and combined with the calibration coefficient of the annular bearing substrate, the eccentricity of the resultant force vector is calculated.

[0113] Specifically, the vector reconstruction unit first obtains the azimuth angles relative to each sensing point on the annular support substrate from the dynamic feature vector. Corresponding force amplitude sequence This sequence reflects the intensity of the load force at discrete points along the circumference of the ring structure within the current sampling period. To obtain a continuous description of the force field, the system performs a sequence of force amplitudes. Nonlinear curve fitting is performed to construct the circumferential force distribution curve. In this embodiment, the curve fitting uses the least squares criterion to ensure that the fitted curve approximates the actual sampling points in each direction to the greatest extent possible.

[0114] In obtaining continuous circumferential force distribution curves Then, the system determines the direction and magnitude of the resultant force vector by searching for the extreme points of the curve.

[0115] Direction angle of the resultant force vector Determined by finding the phase angle corresponding to the maximum amplitude of the curve:

[0116] ;

[0117] Magnitude of the resultant force vector Then, based on the peak amplitude of the curve Decide:

[0118] ;

[0119] In this process, the peak direction directly indicates the orientation of the load offset relative to the center of the lifting point, while the peak amplitude reflects the degree of stress concentration in that direction.

[0120] Eccentricity represents the physical distance between the line of action of the resultant load and the central axis of the annular bearing substrate. The system extracts this parameter by analyzing the fluctuation amplitude of the circumferential force distribution curve. Let the peak value of the circumferential force distribution curve over the entire circumference be... Valley value The peak-to-valley difference reflects the degree of asymmetry in the stress distribution. The system introduces a physical calibration coefficient for the annular load-bearing matrix. This coefficient, pre-determined through discrete load experiments, is used to establish the mapping relationship between stress distribution non-uniformity and physical eccentricity. (Eccentricity of the resultant force vector) The calculation formula is as follows:

[0121] ;

[0122] In practical calculations, if the load is at the absolute equilibrium center, the force distribution curve tends to be flat. Approaching zero, the corresponding eccentricity It also approaches zero. As the load swing angle increases, the stress concentrates highly on one side, leading to an increase in the peak-to-valley difference, thus allowing for the calculation of an accurate eccentricity value. Through the above steps, the vector reconstruction unit output includes the magnitude... ,direction and eccentricity The pose parameter set.

[0123] The attitude calculation unit is used to calculate the pose parameters of the resultant force vector to obtain the spatial attitude data of the load relative to the hoisting equipment.

[0124] Preferably, in the attitude calculation unit, the step of calculating the pose parameters of the resultant force vector includes:

[0125] The swing angle of the load is calculated based on the eccentricity and the effective lever arm radius of the annular bearing base.

[0126] The direction angle of the resultant force vector is directly used as the swing direction of the load;

[0127] Based on the rate of change of the direction angle of the resultant force vector with time, the oscillation angular velocity of the load is calculated, and the direction angle sequence is filtered.

[0128] The swing angle, swing direction, and swing angular velocity are output as spatial attitude data.

[0129] Specifically, the attitude calculation unit first establishes the geometric correspondence between the load swing angle and the force eccentricity, and then uses this eccentricity... With the preset equivalent hoisting length Establish a spatial geometric mapping model. Based on the right-angled triangle relationship formed by the load's center of mass, the center of the lifting point, and the point of application of the resultant force, determine the load swing angle. The calculation formula is as follows:

[0130] ;

[0131] For the direction of load oscillation, the system directly uses the direction angle of the resultant force vector. Mapped to the swing direction angle of the load in the horizontal projection plane ,Right now This angle indicates the orientation of the load deviating from the vertical central axis.

[0132] To obtain complete dynamic attitude information, the system performs time-domain analysis on the orientation angle sequence to calculate the oscillation angular velocity. Let the orientation angles at adjacent sampling times be... and Then the angular velocity of the load swing Represented as:

[0133] ;

[0134] Because mechanical oscillations during hoisting can cause instantaneous jumps in the orientation angle sequence, the system filters the orientation angle sequence. In this embodiment, a Kalman filter algorithm or a moving average filter algorithm is used to process the calculated orientation angle sequence. and Smoothing optimization is performed to eliminate spurious attitude fluctuations caused by high-frequency noise. Finally, the attitude calculation unit calculates and optimizes the sway angle. Swing direction and angular velocity of oscillation Encapsulated as a spatial attitude data packet.

[0135] The communication output module is used to send the calculated spatial attitude data to the control equipment in real time.

[0136] Preferably, in the communication output module, the communication output module sends data to the UAV flight controller via a CAN bus, RS485, or serial interface.

[0137] Specifically, the communication output module outputs signals including at least: off-center load direction, swing angle, and swing speed. Additionally, it may typically include load weight and eccentricity. These signals are used for UAV flight control, and transmission methods can include CAN, RS485, or a serial interface.

[0138] Example 2

[0139] This embodiment establishes a drone hoisting test platform to compare and test the system described in this invention with existing technical solutions, thereby verifying the attitude calculation accuracy and anti-interference capability of this system in complex environments. A simplified flowchart of the system implementation is shown below. Figure 3 As shown. This embodiment is designed as follows:

[0140] Test platform and comparison objects

[0141] Test platform: a hexacopter drone, with a 2kg payload suspended below by a flexible sling. The drone hovers at a height of 5m, and a preset external disturbance is applied, such as a horizontal pulse wind load or rapid ascent and descent of the fuselage.

[0142] Comparison of options:

[0143] Control group 1 (traditional discrete four-point strain): Four strain sensors with a cross-shaped distribution are installed at the suspension point. The eccentricity is estimated only by the amplitude of the four points, and there is no reconstruction of the full circumference stress field.

[0144] Control group 2 (visual scheme): High-speed cameras are deployed on the ground, along with markers at the load end, and the swing angle is obtained through image processing (as a reference true value, but not involved in the closed loop).

[0145] Experimental group: The system of this invention (ring-shaped bearing substrate, 8 circumferential uniform strain gauges, principal component analysis for denoising, Fourier series fitting, vector reconstruction and attitude calculation).

[0146] Test conditions and results

[0147] Condition 1: Strong light + smoke environment (simulating outdoor backlight and dust)

[0148] In control group 2, marker recognition was frequently lost under backlight or smoke conditions, and attitude data showed significant discrepancies. The experimental group, unaffected by light or smoke, maintained continuous and uninterrupted data output, with a stable attitude calculation frequency.

[0149] Operating Condition 2: Emergency Stop and Acceleration / Deceleration of the UAV (Simulated Flight Disturbance)

[0150] The drone suddenly stopped forward with an acceleration of 2 m / s², resulting in a maximum sway angle of approximately 25°. Control group 1 could only output a coarse eccentricity direction, unable to identify the sway speed, and suffered from significant eccentricity noise error due to common-mode interference from drone fuselage vibration. The experimental group effectively isolated common-mode vibration through principal component analysis, significantly reducing the eccentricity noise error, and outputting the sway angular velocity in real time, providing a feedforward compensation signal for flight control.

[0151] Operating Condition 3: Indoor GPS Denial Environment

[0152] Control group 2 failed under low light conditions; control group 1 could still work, but could not calculate the continuous swing direction. The experimental group could still output the 360° continuous swing direction under conditions without GPS and without vision, and the swing angle calculation error was consistent with the outdoor environment.

[0153] Table 1 Comparison of Performance Quantification in Comparative Tests

[0154] Resistance to light / smoke interference good Difference good Electromagnetic interference / GPS denial good Difference good Swing angle calculation error ±5° ±1° (normal light) ±1.2° Oscillating angular velocity output none (30ms delay) (5ms delay) Circumferential resolution 90° (four o'clock) continuous Continuous (1°) Common-mode vibration suppression capability weak not applicable powerful

[0155] The comparative experimental data are shown in Table 1. The comparative test data in the table can prove that the system described in this invention can achieve high-precision and high-interference load attitude calculation without relying on vision, GPS or load-side sensors. It is especially suitable for UAV hoisting scenarios under vision failure, GPS shielding and high dynamic disturbance, and significantly improves the system's environmental adaptability and control robustness.

[0156] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point, characterized in that, include: The ring-shaped load-bearing base is used to connect the hoisting equipment and the hoisted load and to transmit the load force; A full-circumference force sensing unit is set in the circumference of the ring-shaped bearing base to acquire the original force signals of the suspension point structure in different circumferential directions, so as to form a circumferential stress field reflecting the load stress state. The signal preprocessing unit is used to synchronously acquire and process the original force signals from all directions, and extract the dynamic feature vector of the circumferential stress field. The vector reconstruction unit is used to reconstruct the pose parameters of the resultant force vector based on the amplitude distribution law of the feature vector in the circumferential direction, wherein the pose parameters include magnitude, direction and eccentricity. The attitude calculation unit is used to calculate the pose parameters of the resultant force vector to obtain the spatial attitude data of the load relative to the hoisting equipment. The communication output module is used to send the calculated spatial attitude data to the control equipment in real time.

2. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 1, characterized in that, In the full-circumference force sensing unit, the steps for forming a circumferential stress field reflecting the load stress state include: By using discrete sensing points arranged circumferentially along the ring-shaped bearing substrate, the original force signals from different directions are acquired synchronously. Based on the structural geometric parameters and stiffness distribution characteristics of the ring-shaped bearing substrate, the original force signals in each direction are compensated and corrected to eliminate the strain response differences caused by the non-circular structure. The corrected force signal is used as the sampled value, and a curve is fitted using a generalized periodic function containing higher-order harmonic components to reconstruct a continuous closed-loop circumferential stress distribution field in the spatial dimension.

3. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 1, characterized in that, The steps for synchronously acquiring and consistent processing of the original force signals from all directions in the signal preprocessing unit include: In the unloaded state of hoisting, the static output values ​​of sensing points at different orientations are acquired synchronously as the zero-point reference of the original force signal; By using the preset gain deviation coefficient between each sensing point, the original force signal acquired dynamically is corrected so that the amplitude changes of the signal in each direction tend to be consistent under the same force change. Common-mode noise is eliminated for the original force signals in all directions, and the signals are strictly synchronized and aligned in the time dimension to eliminate the phase difference in the dynamic force process.

4. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 3, characterized in that, The steps for common-mode noise cancellation of the original force signals in each direction include: The original force signals from different directions are collected to form a multi-dimensional observation matrix; Principal component analysis was performed on the multidimensional observation matrix, and the principal component with the largest eigenvalue was extracted as the global common mode component. The signal after eliminating common-mode noise is obtained by subtracting the projection of the global common-mode component onto each channel from the original force signals in each direction.

5. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 1, characterized in that, In the signal preprocessing unit, the step of extracting the dynamic feature vector of the circumferential stress field includes: Calculate the rate of change of sensor signals in different orientations over time to obtain dynamic incremental data reflecting the load swing trend; Arrange the original signals from each direction and the dynamic incremental data according to the circumferential angular position to form a spatial vector group containing the circumferential angle, force amplitude and amplitude change rate; From the spatial vector set, common interference components in different directions are removed, and the dominant feature vector that characterizes the change in the eccentric intensity and direction of the circumferential stress field is extracted as the dynamic feature vector.

6. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 5, characterized in that, The steps for removing common interference components from different orientations from the spatial vector set include: Obtain the spatial vector distribution of the load under a preset reference state as a dynamic background reference; The spatial vector group is subjected to point-by-point vector difference with the dynamic background reference to suppress static features caused by environmental factors or structural weight. Calculate the correlation coefficient between each signal and the mean of all signals, and remove global pulsation components whose correlation coefficient is greater than a preset threshold.

7. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point as described in claim 1, characterized in that, In the vector reconstruction unit, the steps for reconstructing the pose parameters of the resultant force vector include: Obtain the force amplitude sequence corresponding to the dynamic feature vector in different directions in the circumferential direction; Curve fitting is performed on the force amplitude sequence to obtain the circumferential force distribution curve; The direction of the peak value of the force distribution curve is taken as the direction of the resultant force vector, and the magnitude of the peak value of the force distribution curve is taken as the magnitude of the resultant force vector. Based on the difference between the peak and valley values ​​of the force distribution curve, and combined with the calibration coefficient of the annular bearing substrate, the eccentricity of the resultant force vector is calculated.

8. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point according to claim 1, characterized in that, The steps for calculating the pose parameters of the resultant force vector in the attitude calculation unit include: The swing angle of the load is calculated based on the eccentricity and the effective lever arm radius of the annular bearing base. The direction angle of the resultant force vector is directly used as the swing direction of the load; Based on the rate of change of the direction angle of the resultant force vector with time, the oscillation angular velocity of the load is calculated, and the direction angle sequence is filtered. The swing angle, swing direction, and swing angular velocity are output as spatial attitude data.

9. The UAV hoisting load attitude calculation system based on the full circumference force distribution of the hoisting point according to claim 1, characterized in that, In the communication output module, the communication output module sends data to the UAV flight controller via CAN bus, RS485, or serial interface.

Images