High-voltage line unmanned aerial vehicle hoisting control method

Abstract

This invention relates to the field of UAV flight control and power transmission line operation technology, and discloses a UAV hoisting control method for high-voltage power lines. This method utilizes a full-tensor magnetic field gradient measurement array, an airborne inertial measurement unit (IMU), and a suspension-point IMU to calculate the UAV's position and yaw state relative to the high-voltage conductor using an analytical model of the high-voltage line magnetic field and nonlinear least squares method. Furthermore, it combines error state extended Kalman filtering, dual IMU yaw angle calculation, a disturbance observer, and dynamic electromagnetic anomaly fuse protection to achieve anti-sway control during hoisting. This invention is beneficial for improving the reliability of state perception, control stability, and safety of UAV hoisting operations in environments with strong near-field electromagnetic interference from high-voltage power lines.

Description

Technical Field

[0001] This invention relates to the field of drone flight control and power transmission line operation technology, and in particular to a method for controlling the hoisting of high-voltage power lines by drones. Background Technology

[0002] With the increasing demand for power transmission line inspection, live-line maintenance, and material delivery operations, using multi-rotor drones carrying flexible slings to perform lifting operations has become an important technical approach in power operation and maintenance. Regarding this type of "drone-suspended load" system, US Patent 9422139B1 discloses that the suspended load, in a swaying state, can have a coupling effect on the aircraft body, thereby affecting flight attitude stability and lifting safety. Therefore, how to balance trajectory tracking and load anti-sway during lifting has become a technical problem of ongoing concern in this field.

[0003] In the prior art, Chinese patent application CN116203981A discloses a trajectory tracking control method for quadrotor UAVs based on disturbance observation. This scheme achieves UAV trajectory tracking by establishing a quadrotor dynamic model and combining disturbance observation with sliding mode control. Furthermore, Chinese patent CN102736632B discloses an electric field differential obstacle avoidance technology for UAVs inspecting live wires, and US patent US9964658B2 discloses a technical approach for electric line detection, avoidance, and navigation using electromagnetic fields. Therefore, it is evident that existing technologies have already studied UAV flight control, line sensing, and operations near live wires.

[0004] However, the aforementioned existing technologies are mainly geared towards conventional quadcopter trajectory control, line detection, or obstacle avoidance scenarios. Their state perception typically still relies on GPS, magnetometers, visual sensors, or electric field strength thresholds for judgment, making it difficult to simultaneously meet the requirements of continuous and stable acquisition of the UAV's position, attitude, and load swing angle in near-field hoisting operations on high-voltage transmission lines. Especially in the near-field area of ​​high-voltage transmission lines, there are strong alternating electromagnetic fields, corona winds, and complex aerodynamic disturbances. Traditional state perception links based on GPS, magnetometers, and vision are easily interfered with, making it difficult to stably acquire the UAV's position relative to the high-voltage transmission line, and making it difficult to accurately perceive the load swing angle in real time.

[0005] Furthermore, regarding the electromagnetic interference problem near high-voltage power lines, international patent WO2019113424A1 discloses a technical approach to suppress interference through conductive shielding structures. While such solutions can reduce the impact of electromagnetic radiation to some extent, they often lead to problems such as increased structural weight, reduced payload, shortened flight range, and limited communication links. Moreover, their suppression of power frequency alternating magnetic field interference from high-voltage power lines remains limited. Therefore, existing technologies still suffer from insufficient reliability of state perception, low stability of anti-sway control, and difficulty in simultaneously ensuring safety during UAV hoisting operations under strong near-field electromagnetic interference conditions near high-voltage power lines. Summary of the Invention

[0006] Technical issues

[0007] The technical problem to be solved by this invention is to address the issue that strong electromagnetic interference in the near-field region of high-voltage transmission lines causes traditional GPS, magnetometers, and visual perception methods to easily fail, making it difficult for UAV hoisting systems to stably obtain relative position and load swing angle status, and making it difficult to reliably execute anti-sway control and trajectory tracking. Therefore, this invention proposes a UAV hoisting control method for high-voltage lines.

[0008] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0009] A control method for hoisting high-voltage power lines using an unmanned aerial vehicle (UAV) is proposed. The method includes: acquiring spatial magnetic field gradient tensor and multi-source kinematic data; using a full-tensor magnetic field gradient measurement array fixedly mounted on the frame of the multi-rotor UAV to collect real-time three-dimensional spatial gradient tensor data of the AC magnetic field around the high-voltage power line; simultaneously acquiring the three-axis acceleration and angular velocity data of the airborne inertial measurement unit and the three-axis acceleration data of the suspension point output by the suspension point inertial measurement unit; performing anti-electromagnetic interference self-sensing state estimation; substituting the three-dimensional spatial gradient tensor data into a pre-established analytical model of the high-voltage power line magnetic field distribution; calculating the relative position vector and relative yaw angle of the UAV's center of mass relative to the high-voltage power line using a nonlinear least squares method; and using the calculation results as position observation information during the observation update stage of the error state extended Kalman filter. Simultaneously, the three-axis acceleration data of the airborne inertial measurement unit are also acquired. The difference between the acceleration data and the triaxial acceleration data at the suspension point is used, combined with kinematic constraints, to calculate the spatial swing angle and angular velocity of the hoisted load; nonlinear anti-sway and trajectory tracking control laws are calculated, and an underactuated system model containing the dynamic coupling terms of the UAV and the hoisted load is constructed based on Lagrange dynamics. A sliding surface is defined as a linear combination of the UAV position tracking error and the spatial swing angle, and the system virtual control thrust vector used to simultaneously suppress load swing and track the desired spatial trajectory is calculated in combination with the external lumped disturbance estimate output by the disturbance observer; hoisting command allocation and drive control are executed, and the system virtual control thrust vector is decoupled into the desired total thrust and the desired attitude rotation matrix, and mapped to the speed control commands of each rotor motor of the multi-rotor UAV through the hybrid control matrix, so as to drive the multi-rotor UAV to perform anti-sway hoisting operations in the space around the high-voltage line.

[0010] Furthermore, the full tensor magnetic field gradient measurement array consists of at least four three-axis fluxgate sensors. The at least four three-axis fluxgate sensors are fixedly arranged in a cross or star topology on the horizontal plane of the frame, and the baseline distance between them is fixed. By performing spatial difference on the measurement values ​​of each sensor, a three-dimensional spatial gradient tensor characterizing the spatial change rate of the magnetic field is obtained, so as to eliminate the DC component of the geomagnetic field background and the low-frequency electromagnetic interference generated by the airborne motor.

[0011] Furthermore, the process of calculating the relative position vector and relative yaw angle of the UAV's centroid relative to the high-voltage power line using the nonlinear least squares method includes: constructing a theoretical magnetic field gradient tensor model of the high-voltage power line based on the Biot-Savart law, and using the deviation between the three-dimensional spatial gradient tensor data and the theoretical magnetic field gradient tensor model as the objective function, and iteratively minimizing the solution using the Levenberg-Marquardt algorithm. When the iteration step size is less than a preset convergence threshold, the relative position vector and relative yaw angle of the current iteration are output as the position observation information.

[0012] Furthermore, the process of calculating the spatial swing angle and oscillation angular velocity of the hoisted load includes: establishing the kinematic constraint relationship between the suspension point acceleration measured by the suspension point inertial measurement unit, the body acceleration measured by the airborne inertial measurement unit, the body angular velocity, the body angular acceleration, and the fixed position vector from the body's center of mass to the suspension point; obtaining the tilt vector of the flexible sling in space by extracting the horizontal projection component of the suspension point acceleration in the gravity coordinate system; and obtaining the oscillation angular velocity by performing time differentiation on the tilt vector.

[0013] Furthermore, the underactuated system model includes load coupling terms related to the mass of the UAV, the mass of the hoisted load, the length of the flexible sling, the angular acceleration of the spatial swing angle, and the swing angular velocity. The sliding surface is composed of the position tracking error, the rate of change of the position tracking error, and the spatial swing angle, so that the UAV applies damping constraints to the load swing during trajectory tracking.

[0014] Furthermore, the disturbance observer estimates the external lumped disturbances, including corona wind disturbances and aerodynamic disturbances, in real time based on the system's virtual control thrust vector, UAV speed state, and dynamic model, and feeds the estimated external lumped disturbance values ​​forward to compensate the nonlinear anti-sway and trajectory tracking control law.

[0015] Furthermore, the process of executing hoisting command allocation and drive control also includes dynamic electromagnetic anomaly fuse protection logic: calculating the Frobenius norm of the magnetic field gradient tensor matrix output by the full tensor magnetic field gradient measurement array in real time, and monitoring the difference in the trace of the state covariance matrix output by the error state extended Kalman filter in adjacent time steps; when the Frobenius norm is greater than a preset breakdown warning threshold or the difference in the trace of the state covariance matrix is ​​greater than a preset divergence threshold, triggering an electromagnetic anomaly fuse command, resetting the desired attitude rotation matrix to an identity matrix and locking the direction of the system virtual control thrust vector to the opposite direction of gravity, controlling the multi-rotor UAV to climb vertically with maximum available thrust until the Frobenius norm is less than the safety recovery threshold.

[0016] Beneficial effects: Compared with existing UAV hoisting control schemes that rely on GPS, magnetometers, or visual perception, this invention sets up a full-tensor magnetic field gradient measurement array on a multi-rotor UAV, and substitutes the collected three-dimensional spatial gradient tensor data of the AC magnetic field around the high-voltage line into the high-voltage line magnetic field distribution analytical model for nonlinear least squares solution. Then, it combines error state extended Kalman filtering for state update, so that the electromagnetic field around the high-voltage line, which is originally prone to interference, is transformed into relative positioning observation information. This is beneficial for maintaining stable perception of the UAV's position relative to the high-voltage line in the context of strong electromagnetic interference in the near field of the high-voltage line.

[0017] This invention further calculates the spatial swing angle and oscillation angular velocity of the hoisted load by differential kinematic constraints between the airborne inertial measurement unit and the suspension point inertial measurement unit. It can obtain load state information for anti-sway control without relying on external visual perception links that are susceptible to background texture, lighting changes and electromagnetic environment, thereby improving the reliability of load swing state perception in high-voltage line near-field hoisting scenarios.

[0018] This invention also constructs an underactuated system model that includes the coupling term of the dynamics of the UAV and the load being hoisted. It combines the sliding surface composed of the position tracking error and the spatial swing angle with the external lumped disturbance estimate output by the disturbance observer to calculate the control quantity. This enables trajectory tracking and load anti-swing to be achieved in a coordinated manner under a unified control framework, which is beneficial to improving the control robustness and operational stability when facing corona wind disturbances and aerodynamic disturbances during high-voltage line near-field hoisting.

[0019] Furthermore, by setting up dynamic electromagnetic anomaly fuse protection logic based on the Frobenius norm of the magnetic field gradient tensor matrix and the trace change of the state covariance matrix, this invention can promptly switch to vertical climbing escape control under abnormal electromagnetic conditions, which is beneficial to improving the safety assurance capability in high-voltage line near-field hoisting operations. Attached Figure Description

[0020] Figure 1 This is a system physical architecture diagram for the high-voltage line drone hoisting scenario in this invention;

[0021] Figure 2 The overall control flow logic diagram of the anti-sway and anti-interference control method of the present invention includes four stages: magnetic field signal processing, ES-EKF state estimation, SMC anti-sway control, and motor mixed control.

[0022] Figure 3 A diagram defining the multi-rigid-body topology and kinematic coordinate system of an unmanned aerial vehicle (UAV) and a flexible suspended load system is provided, with the inertial frame, mechanical frame, and high-voltage line reference frame labeled.

[0023] Figure 4 This is a schematic diagram of the Levenberg-Marquardt nonlinear optimization localization solution based on the full tensor magnetic field gradient tensor.

[0024] Figure 5 The diagram shows the internal control loop of the proposed sliding mode controller (SMC) and disturbance observer (DOB). Detailed Implementation

[0025] To enable those skilled in the art to more clearly understand the technical solution of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Various modifications and substitutions can be made by those skilled in the art without departing from the concept of the present invention, and all such modifications and substitutions should fall within the scope of protection of the present invention.

[0026] Example 1:

[0027] like Figure 1 As shown, the high-voltage power line UAV hoisting system in this embodiment includes a multi-rotor UAV, a flexible sling, a hoisting load, an airborne inertial measurement unit, a suspension point inertial measurement unit, a full tensor magnetic field gradient measurement array, and a processor and memory for executing control algorithms. The full tensor magnetic field gradient measurement array is fixedly mounted on the frame of the multi-rotor UAV and is used to collect three-dimensional spatial gradient information of the AC magnetic field around the high-voltage power line. The airborne inertial measurement unit is used to collect the three-axis acceleration and three-axis angular velocity of the UAV body. The suspension point inertial measurement unit is located at the connection point between the flexible sling and the UAV and is used to collect suspension point acceleration information. The processor is used to execute near-field relative positioning of the high-voltage power line, swing angle calculation, anti-swing control, and abnormal fuse protection logic.

[0028] like Figure 2As shown, the high-voltage power line UAV hoisting control method in this embodiment is as follows: Acquire spatial magnetic field gradient tensor and multi-source kinematic data; real-time acquisition of three-dimensional spatial gradient tensor data of the AC magnetic field around the high-voltage power line using the full-tensor magnetic field gradient measurement array fixedly installed on the multi-rotor UAV frame; synchronous acquisition of the body's three-axis acceleration and three-axis angular velocity data output by the airborne inertial measurement unit, and the suspension point's three-axis acceleration data output by the suspension point inertial measurement unit; Perform anti-electromagnetic interference self-sensing state estimation; substitute the three-dimensional spatial gradient tensor data into a pre-established analytical model of the high-voltage power line magnetic field distribution; calculate the relative position vector and relative yaw angle of the UAV's center of mass relative to the high-voltage power line using the nonlinear least squares method; and use the calculation results as position observations during the error state extended Kalman filter observation update stage. The system simultaneously uses the difference between the three-axis acceleration data of the aircraft body and the three-axis acceleration data of the suspension point, combined with kinematic constraints, to calculate the spatial swing angle and angular velocity of the hoisted load; calculates the nonlinear anti-sway and trajectory tracking control law, constructs an underactuated system model based on Lagrange dynamics that includes the dynamic coupling terms of the UAV and the hoisted load, defines a sliding surface composed of a linear combination of the UAV position tracking error and the spatial swing angle, and calculates the system virtual control thrust vector used to simultaneously suppress load swing and track the desired spatial trajectory by combining the external lumped disturbance estimate output by the disturbance observer; executes hoisting command allocation and drive control, decouples the system virtual control thrust vector into the desired total thrust and the desired attitude rotation matrix, and maps it to the speed control command of each rotor motor of the multi-rotor UAV through the hybrid control matrix. The high-voltage line UAV hoisting control method of this embodiment can be divided into four stages: data acquisition, state estimation, anti-sway control, and command execution.

[0029] During the data acquisition phase, the processor synchronously receives output data from the full tensor magnetic field gradient measurement array, the airborne inertial measurement unit, and the suspension point inertial measurement unit.

[0030] During the state estimation phase, the processor uses magnetic field gradient tensor data and an analytical model of the high-voltage line magnetic field distribution to solve for the relative position vector and relative yaw angle of the UAV relative to the high-voltage line, and uses the solution results as the observation information for the error state extended Kalman filter; at the same time, the spatial swing angle and oscillation angular velocity of the hoisted load are calculated based on the differential kinematic relationship between the airborne inertial measurement unit and the suspension point inertial measurement unit.

[0031] During the anti-sway control phase, the processor establishes trajectory tracking and anti-sway control laws based on the coupled dynamics model of the UAV and the hoisted load, and compensates for the control quantity by combining the external lumped disturbance estimated in real time by the disturbance observer.

[0032] During the instruction execution phase, the processor converts the system virtual control thrust vector obtained from the control law into the desired total thrust and desired attitude rotation matrix, and then maps it into the speed control commands of each rotor motor of the multi-rotor UAV through the hybrid control matrix, so as to drive the multi-rotor UAV to complete the hoisting task around the high-voltage line.

[0033] In this embodiment, the full tensor magnetic field gradient measurement array consists of at least four triaxial fluxgate sensors. These sensors are arranged in a cross or star topology on the horizontal plane of the frame, and a spatial difference relationship is established through a fixed baseline distance. By spatially differentiating the magnetic field vectors output by each sensor, a gradient tensor characterizing the spatial rate of change of the magnetic field can be obtained, thereby reducing the impact of the background DC component of the geomagnetic field and the low-frequency electromagnetic interference from the UAV itself on the positioning solution.

[0034] The process of calculating the relative position vector and relative yaw angle of the UAV's centroid relative to the high-voltage power line using the nonlinear least squares method includes: constructing a theoretical magnetic field gradient tensor model of the high-voltage power line based on the Biot-Savart law; using the deviation between the three-dimensional spatial gradient tensor data and the theoretical magnetic field gradient tensor model as the objective function; iteratively minimizing the deviation using the Levenberg-Marquardt algorithm; and outputting the relative position vector and relative yaw angle of the current iteration as the position observation information when the iteration step size is less than a preset convergence threshold.

[0035] The process of calculating the spatial swing angle and oscillation angular velocity of the hoisted load includes: establishing the kinematic constraint relationship between the suspension point acceleration measured by the suspension point inertial measurement unit, the body acceleration measured by the airborne inertial measurement unit, the body angular velocity, the body angular acceleration, and the fixed position vector from the body's center of mass to the suspension point; obtaining the tilt vector of the flexible sling in space by extracting the horizontal projection component of the suspension point acceleration in the gravity coordinate system; and obtaining the oscillation angular velocity by taking the time derivative of the tilt vector.

[0036] The underactuated system model includes load coupling terms related to the mass of the UAV, the mass of the hoisted load, the length of the flexible sling, the angular acceleration of the spatial swing angle, and the swing angular velocity. The sliding surface is composed of the position tracking error, the rate of change of the position tracking error, and the spatial swing angle, so as to apply damping constraints to the load swing during the UAV's trajectory tracking process.

[0037] The disturbance observer estimates the external lumped disturbances, including corona wind disturbances and aerodynamic disturbances, in real time based on the system's virtual control thrust vector, UAV speed state, and dynamic model, and feeds the estimated external lumped disturbance values ​​forward to compensate the nonlinear anti-sway and trajectory tracking control law.

[0038] The process of executing hoisting command allocation and drive control also includes dynamic electromagnetic anomaly fuse protection logic: real-time calculation of the Frobenius norm of the magnetic field gradient tensor matrix output by the full tensor magnetic field gradient measurement array, and monitoring the difference in the trace of the state covariance matrix output by the error state extended Kalman filter in adjacent time steps; when the Frobenius norm is greater than a preset breakdown warning threshold or the difference in the trace of the state covariance matrix is ​​greater than a preset divergence threshold, an electromagnetic anomaly fuse command is triggered, the desired attitude rotation matrix is ​​reset to an identity matrix and the direction of the system virtual control thrust vector is locked to the opposite direction of gravity, so as to control the multi-rotor UAV to climb vertically with the maximum available thrust until the Frobenius norm is less than the safety recovery threshold.

[0039] like Figure 3 As shown, to facilitate the description of the relative positional relationship between the UAV and the high-voltage power line, this embodiment defines the Earth inertial coordinate system, the UAV's frame of reference, and the high-voltage power line's reference frame. The relative position vector of the UAV relative to the high-voltage power line is denoted as... The relative yaw angle is denoted as .

[0040] like Figure 4 As shown, after receiving the magnetic field gradient tensor data output by the full tensor magnetic field gradient measurement array, the processor substitutes it into the analytical model of the high-voltage line magnetic field distribution and constructs the following objective function:

[0041]

[0042] in, Indicates the first The magnetic field gradient tensor measured by each sensor This represents the gradient tensor calculated from the theoretical magnetic field model of high-voltage lines. This indicates the number of sensors involved in the solution. The processor uses the Levenberg-Marquardt algorithm to iteratively minimize the objective function, outputting the relative position vector of the current iteration when the iteration step size is less than a preset convergence threshold. With relative yaw angle .

[0043] After obtaining the relative position calculation results, the processor uses them as the observation input for the error state extended Kalman filter, and the body acceleration and angular velocity output by the airborne inertial measurement unit as the prediction input, to achieve continuous estimation of the UAV's position, velocity, and attitude. Thus, the AC magnetic field around the high-voltage power line is no longer merely considered an interference source, but is converted into observational information for near-field relative positioning of the high-voltage line.

[0044] like Figure 3As shown, the load is suspended below the multi-rotor UAV via a flexible sling. To obtain the spatial swing state of the load, the processor establishes kinematic constraints between the suspension point acceleration, the body acceleration, the body angular velocity, and the vector from the body's center of mass to the fixed suspension point:

[0045]

[0046] in, This represents the suspension point acceleration measured by the suspension point inertial measurement unit. This represents the airframe acceleration measured by the airborne inertial measurement unit. Indicates the angular velocity of the machine body. This represents the fixed position vector from the body's center of mass to the suspension point. The processor, based on the aforementioned constraints and by extracting the horizontal projection component of the suspension point acceleration in the gravity coordinate system, obtains the cable inclination vector. And through the Find the time derivative to obtain the angular velocity of the oscillation.

[0047] like Figure 5 As shown, after obtaining the UAV's position state and the sling load's swing angle state, the processor constructs an underactuated dynamic model that includes the sling load coupling term and defines the sliding surface.

[0048]

[0049] in, This indicates the drone's position tracking error. and This represents the sliding mode gain matrix. Based on the sliding mode surface, the processor calculates the system virtual control thrust vector used to simultaneously suppress load oscillations and track the desired trajectory. .

[0050] Furthermore, to compensate for corona wind disturbances and aerodynamic disturbances, the processor establishes a disturbance observer and calculates the estimated external lumped disturbance in real time according to the following update relationship:

[0051]

[0052]

[0053] in, Indicates the internal state of the observer. Represents the observation gain matrix. Indicates the quality of the drone. Indicates the mass of the load being hoisted. This represents the estimated external lumped disturbance. The processor will then... Feedforward compensation is incorporated into the control law to improve the robustness of the system to external disturbances in the near-field environment of high-voltage lines.

[0054] In this embodiment, the processor then virtually controls the thrust vector of the system. Decoupling to desired total thrust and desired pose rotation matrix And the desired total thrust is controlled by a hybrid control matrix. and desired pose rotation matrix The commands are mapped to the speed control commands of each rotor motor, thereby achieving coordinated control of trajectory tracking and load anti-sway.

[0055] like Figure 5 As shown, in addition to the normal control loop, this embodiment also includes dynamic electromagnetic anomaly fuse protection logic. The processor calculates the Frobenius norm of the magnetic field gradient tensor matrix output by the full tensor magnetic field gradient measurement array in real time and monitors the change of the trace of the state covariance matrix output by the error state extended Kalman filter in adjacent time steps. When the Frobenius norm of the magnetic field gradient tensor matrix is ​​greater than a preset breakdown warning threshold, or the difference in the traces of the state covariance matrix is ​​greater than a preset divergence threshold, the processor triggers an electromagnetic anomaly fuse command.

[0056] After triggering the electromagnetic anomaly fuse command, the processor resets the desired attitude rotation matrix to an identity matrix and locks the direction of the system's virtual control thrust vector to the opposite direction of gravity. It then uses maximum available thrust to control the multi-rotor UAV's vertical climb until the Frobenius norm of the magnetic field gradient tensor matrix falls below the safe recovery threshold, at which point the normal anti-sway control process resumes. This setting reduces the risk of collision with high-voltage lines under abnormal near-field electromagnetic conditions.

[0057] This invention can also be implemented in the form of a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to sequentially complete... Figure 2 The system includes multi-source data acquisition, magnetic field gradient positioning, swing angle calculation, nonlinear anti-swing control, mixed control allocation, and abnormal fuse protection.

Claims

1. A hoisting method for high-voltage power line unmanned aerial vehicle (UAV) hoisting scenarios, characterized in that, The method includes: acquiring spatial magnetic field gradient tensor and multi-source kinematic data; real-time acquisition of three-dimensional spatial gradient tensor data of the AC magnetic field around the high-voltage line using a full-tensor magnetic field gradient measurement array fixedly mounted on the frame of the multi-rotor UAV; simultaneously acquiring the three-axis acceleration and three-axis angular velocity data of the airborne inertial measurement unit and the three-axis acceleration data of the suspension point output by the suspension point inertial measurement unit; performing anti-electromagnetic interference self-sensing state estimation; substituting the three-dimensional spatial gradient tensor data into a pre-established analytical model of the high-voltage line magnetic field distribution; calculating the relative position vector and relative yaw angle of the UAV's center of mass relative to the high-voltage line using a nonlinear least squares method; and using the calculation results as position observation information during the observation update stage of the error state extended Kalman filter, while utilizing... The difference between the three-axis acceleration data of the airframe and the three-axis acceleration data of the suspension point is used to calculate the spatial swing angle and oscillation angular velocity of the hoisted load by combining kinematic constraints; nonlinear anti-sway and trajectory tracking control laws are calculated, and an underactuated system model containing the dynamic coupling terms of the UAV and the hoisted load is constructed based on Lagrange dynamics. A sliding surface is defined by a linear combination of the UAV position tracking error and the spatial swing angle, and the system virtual control thrust vector used to simultaneously suppress load sway and track the desired spatial trajectory is calculated by combining the external lumped disturbance estimate output by the disturbance observer; hoisting command allocation and drive control are executed, and the system virtual control thrust vector is decoupled into the desired total thrust and the desired attitude rotation matrix, and mapped to the speed control commands of each rotor motor of the multi-rotor UAV through the hybrid control matrix.

2. The hoisting method according to claim 1, characterized in that, The full tensor magnetic field gradient measurement array consists of at least four triaxial fluxgate sensors. The at least four triaxial fluxgate sensors are fixedly arranged in a cross or star topology on the horizontal plane of the frame, and the baseline distance between them is fixed. The three-dimensional spatial gradient tensor characterizing the spatial rate of change of the magnetic field is obtained by spatially differentiating the measurement values ​​of each sensor.

3. The hoisting method according to claim 1, characterized in that, The process of calculating the relative position vector and relative yaw angle of the UAV's centroid relative to the high-voltage power line using the nonlinear least squares method includes: constructing a theoretical magnetic field gradient tensor model of the high-voltage power line based on the Biot-Savart law; using the deviation between the three-dimensional spatial gradient tensor data and the theoretical magnetic field gradient tensor model as the objective function; iteratively minimizing the deviation using the Levenberg-Marquardt algorithm; and outputting the relative position vector and relative yaw angle of the current iteration as the position observation information when the iteration step size is less than a preset convergence threshold.

4. The hoisting method according to claim 1, characterized in that, The process of calculating the spatial swing angle and oscillation angular velocity of the hoisted load includes: establishing the kinematic constraint relationship between the suspension point acceleration measured by the suspension point inertial measurement unit, the body acceleration measured by the airborne inertial measurement unit, the body angular velocity, the body angular acceleration, and the fixed position vector from the body's center of mass to the suspension point; obtaining the tilt vector of the flexible sling in space by extracting the horizontal projection component of the suspension point acceleration in the gravity coordinate system; and obtaining the oscillation angular velocity by performing time differentiation on the tilt vector.

5. The hoisting method according to claim 1, characterized in that, The underactuated system model includes load coupling terms related to the mass of the UAV, the mass of the hoisted load, the length of the flexible sling, the angular acceleration of the spatial swing angle, and the angular velocity of the swing. The sliding surface is composed of the position tracking error, the rate of change of the position tracking error, and the spatial swing angle.

6. The hoisting method according to claim 1, characterized in that, The disturbance observer estimates the external lumped disturbances, including corona wind disturbances and aerodynamic disturbances, in real time based on the system's virtual control thrust vector, UAV speed state, and dynamic model, and feeds the estimated external lumped disturbance values ​​forward to compensate the nonlinear anti-sway and trajectory tracking control law.

7. The hoisting method according to claim 1, characterized in that, The process of executing hoisting command allocation and drive control also includes dynamic electromagnetic anomaly fuse protection logic: real-time calculation of the Frobenius norm of the magnetic field gradient tensor matrix output by the full tensor magnetic field gradient measurement array, and monitoring the difference in the trace of the state covariance matrix output by the error state extended Kalman filter in adjacent time steps; when the Frobenius norm is greater than a preset breakdown warning threshold or the difference in the trace of the state covariance matrix is ​​greater than a preset divergence threshold, an electromagnetic anomaly fuse command is triggered, the desired attitude rotation matrix is ​​reset to an identity matrix and the direction of the system virtual control thrust vector is locked to the opposite direction of gravity, so as to control the multi-rotor UAV to climb vertically with the maximum available thrust until the Frobenius norm is less than the safety recovery threshold.

Images