A path planning and optimization method for unmanned aerial vehicle power inspection task

Abstract

The present application relates to the field of intelligent unmanned aerial vehicle manufacturing, and discloses a path planning and optimization method for unmanned aerial vehicle power inspection tasks, comprising: calculating a deviation vector and a tracking deviation vector gradient; converting discrete residual values of an inverter reverse back electromotive force of an electric drive rudder into phase lag characteristic parameters to adjust the width of a sliding time window; implementing discrete integration on the tracking deviation vector gradient in the adjusted sliding time window to output a transient cumulative energy scalar; issuing a trajectory reconstruction trigger instruction when the integral energy threshold is crossed for 5 consecutive periods, converting the tracking deviation vector gradient into an attitude angle expected correction increment matrix using a linear feedforward compensation matrix and injecting the axis servo loop, otherwise maintaining regular control, the present application establishes a closed loop between the sliding sampling bandwidth and the rudder state to hedge the physical lag boundary, eliminates loop saturation oscillation and guarantees attitude stability.

Description

Technical Field

[0001] This invention relates to a path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks, belonging to the field of intelligent unmanned aerial vehicle manufacturing technology. Background Technology

[0002] Unmanned aerial vehicle (UAV) power line inspections typically require flying along pre-set inspection routes. Existing path planning and route-keeping methods mostly generate flight paths based on known coordinates or real-time environmental information. The onboard controller then calculates the deviation between the UAV's current position and the target route and uses closed-loop feedback to drive the control mechanism to adjust the flight attitude, keeping the UAV as close to the preset route as possible. In environments with relatively stable airflow and minimal interference, this method can achieve good trajectory tracking and basically meets the automation requirements of inspection operations. The control process essentially transmits the flight deviation to the underlying actuators and then continues to correct the attitude based on the actuators' responses. The control algorithm generates attitude correction commands based on the deviation signal, and the electrically driven servo motors then correct the flight deviation by changing the output deflection angle.

[0003] In power line inspection scenarios, drones often need to fly close to high-voltage transmission towers and power lines. The inspection channel is narrow, and strong electromagnetic interference and unstable airflow may exist simultaneously. Onboard sensors are prone to random jitter noise or short-term data loss. Sudden shear airflow can also cause rapid fluctuations in the drone's attitude within a short period. To maintain flight path tracking, conventional control schemes typically perform frequent position signal filtering and fine-tuning of control parameters with significant gain. However, this approach easily leads to phase lag in the feedback signal and makes it difficult to reflect the discrete residual value of the inverter back electromotive force accumulated during high-frequency adjustment of the electric drive servo motor. As a result, the time scale of the front-end control judgment and the actual response lag of the underlying servo mechanism can easily become misaligned, leading to saturation oscillations in the control loop. This type of control process not only exposes the inherent limitations of the physical configuration of the control mechanism... Due to limitations, the software algorithm of its control method also has shortcomings. For example, Chinese invention patent application CN121979246A discloses an event-triggered preset performance control method for inspection drones for monitoring the condition of transmission line insulators. It uses a logarithmic function to carry out preset performance error transformation. However, this method implicitly relies on the underlying objective property that the tracking error is always strictly limited within the performance boundary. In actual inspection channels with strong electromagnetic interference and sudden shear airflow coupling, the transient step of the drone's state can easily cause the error to instantly cross the preset boundary, causing the logarithmic mapping to fall into mathematical singularity and leading to algorithm collapse. At the same time, it fails to take into account the discrete residual of the inverter reverse electromotive force accumulated under the high-frequency adjustment of the servo circuit, and does not introduce a frequency limiting mechanism for the transient power supply capability of the energy system. It is prone to saturation oscillation and overshoot risk during large-angle trajectory reconstruction.

[0004] Therefore, in the power inspection environment under the combined effects of strong electromagnetic interference and unsteady shear airflow, how to combine the physical state feedback of the actuator and the airborne power supply capability to dynamically adjust the discrete control process and establish a closed-loop attitude adjustment loop that can compensate for the time delay has become a technical problem that needs to be solved. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks, comprising the following steps: Step S101: Collect the absolute space three-dimensional coordinate sequence of the unmanned aerial vehicle, calculate the real-time deviation vector between the absolute space three-dimensional coordinate sequence and the pre-stored benchmark inspection route, and calculate the tracking deviation vector gradient of the current control cycle by the time domain difference of the real-time deviation vector of adjacent control cycles. Step S102: Detect the discrete residual value of the inverter reverse electromotive force generated by the bottom electric drive servo in the previous control cycle and convert it into a phase lag characterization parameter. Adjust the sliding time window width of the current control cycle according to the phase lag characterization parameter so that the sliding time window width is an integer value after rounding and converting the sum of the reference window constant and the phase lag characterization parameter. Step S103: The gradient of the tracking deviation vector is introduced into the sliding time window after the phase lag characterization parameter is adjusted to perform discrete integration calculation, and the transient cumulative energy scalar is generated. Step S104: Compare the transient cumulative energy scalar with the preset integral energy gating threshold. If the transient cumulative energy scalar crosses the integral energy gating threshold within 5 consecutive control cycles, issue a trajectory reconstruction trigger command to activate the path space reconstruction control program. Use the linear feedforward compensation matrix to convert the tracking deviation vector gradient into the attitude angle expectation correction increment matrix and inject it into the axis servo loop. Otherwise, switch to the normal route hold control mode and call the proportional-derivative control algorithm to maintain stable flight within the safe envelope.

[0006] Preferably, step S101 includes the following sub-steps: Step S1011, when the unmanned aerial vehicle encounters local electromagnetic field interference causing intermittent occlusion of spatial deviation data and discrete loss, the airborne inertial control state discrete prediction subroutine is started; Step S1012, the state transition matrix parameters are calculated based on the kinematic multi-order derivative vectors retained before the disconnection; Step S1013, the lost data sequence is filled in by the prediction output of the airborne inertial control state discrete prediction subroutine, and the tracking deviation vector gradient is determined.

[0007] Preferably, in step S104, the integral energy gating threshold is determined by the discrete integral energy criterion; the transient cumulative energy scalar obtained by discretizing and integrating the gradient of the continuously input tracking deviation vector within the sliding time window is used to reflect the high-frequency disturbance energy intensity of the control loop when the UAV is tracking at close range in a narrow inspection channel.

[0008] Preferably, in step S104, the path space reconstruction control program calls the preset adaptive attitude control loop toolbox; the tracking deviation vector gradient is input into the linear feedforward compensation matrix, and the desired attitude angle correction increment matrix is ​​calculated by mapping.

[0009] Preferably, the adaptive attitude control loop toolbox is equipped with a maximum switching frequency limit parameter; the maximum switching frequency limit parameter monotonically constrains the maximum action frequency based on the transient power supply capability of the airborne physical energy system; when the axis servo loop load fluctuates due to large-angle trajectory reconstruction, the system automatically reduces the maximum action switching frequency band of the attitude axis according to the power battery voltage drop trend, limits the overshoot amplitude of the electric drive servo motor under low voltage conditions, suppresses the mechanical resonance overshoot of the axis servo loop, and ensures the action coordination of the motion mechanism under limited airborne computing power and energy constraints.

[0010] Preferably, in step S104, issuing a trajectory reconstruction trigger command to activate the path space reconstruction control program includes the following sub-steps: Step S1041, determining the occurrence of an out-of-bounds discrete event through a discrete threshold arbitration state machine, and activating the path space reconstruction control program; Step S1042, substituting the tracking deviation vector gradient into the linear feedforward compensation matrix, and generating an attitude angle expected correction increment matrix through nonlinear mapping; Step S1043, injecting the attitude angle expected correction increment matrix into the axis servo loop in real time, driving the unmanned aerial vehicle to change its motion attitude to avoid the center of strong electromagnetic interference.

[0011] Preferably, in step S104, switching to the conventional flight path holding control mode includes the following sub-steps: Step S1044, when the transient cumulative energy scalar does not cross the integral energy gating threshold for 5 consecutive control cycles, the discrete threshold arbitration state machine suspends the subsequent spatial deformation matrix calculation; Step S1045, the system automatically switches to the conventional flight path holding control mode by reducing dimensions, calling the proportional-derivative control algorithm to maintain flight path tracking control within the safe envelope range, filtering electromagnetic flypoint noise caused by intermittent occlusion, and avoiding frequent switching of the discrete threshold arbitration state machine between different control commands.

[0012] Preferably, before the unmanned aerial vehicle takes off, the state space prediction operator and the discrete threshold arbitration state machine are initialized through the onboard embedded real-time operating system, a time-sensitive closed-loop support environment is established for the state space prediction operator and the discrete threshold arbitration state machine, and a safety envelope range is established according to the spatial geometric constraint distribution of the inspection channel.

[0013] Preferably, the tracking deviation vector gradient is calculated in real time based on the deviation between the current attitude angle of the UAV and the pre-stored reference inspection route; the deviation includes position deviation, velocity deviation and angle deviation.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. In the path planning of UAV power line inspection mission, by converting the discrete residual value of the inverter reverse electromotive force generated by the electric drive servo into a phase lag characterization parameter, and adjusting the width of the sliding time window accordingly, the front-end sampling rhythm can be matched with the actual response state of the servo. When the UAV experiences a slowdown in servo response due to sudden airflow interference, the system can appropriately widen the sampling judgment time, reduce the sensitivity to short-term positioning jitter, and avoid ordinary noise being misjudged as an abnormal situation requiring route reconstruction. This reduces the saturation oscillation of the control loop caused by continuous high-gain fine-tuning, making the flight attitude adjustment more stable.

[0015] 2. By accumulating and judging the gradient of the tracking deviation vector within the adjusted sliding time window, and using the comparison result of the transient accumulated energy scalar and the integral energy gating threshold as the triggering basis, the system can distinguish between short-term jitter and continuous deviation. Only when the deviation accumulates to a certain extent will the path space reconstruction control program be started; if the triggering condition is not met, the normal flight path holding control mode will be maintained. This can filter out electromagnetic flypoint noise caused by strong electromagnetic interference and intermittent obstruction, and avoid frequent switching of the discrete threshold arbitration state machine between different control commands, which is conducive to maintaining the attitude stability of the UAV during the inspection process.

[0016] 3. By setting a maximum switching frequency limit parameter in the adaptive attitude control loop and constraining it by the power battery's power supply capacity, the system can actively limit the attitude axis action switching frequency and the overshoot amplitude of the electric drive servo when the battery voltage drops or the axis servo loop load fluctuates. This can reduce the risk of servo mechanism overshoot and mechanical resonance under low voltage conditions, enabling the UAV to maintain good motion coordination under limited onboard computing power and energy conditions. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method for planning and optimizing the flight path of a UAV power line inspection according to the present invention; Figure 2 This is a hardware structure diagram of the UAV power inspection control device of the present invention.

[0018] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A method for path planning and optimization of unmanned aerial vehicle (UAV) power line inspection tasks includes the following steps: Step S101: Collect the absolute space three-dimensional coordinate sequence of the unmanned aerial vehicle, calculate the real-time deviation vector between the absolute space three-dimensional coordinate sequence and the pre-stored benchmark inspection route, and calculate the tracking deviation vector gradient of the current control cycle by the time domain difference of the real-time deviation vector of adjacent control cycles. Step S102: Detect the discrete residual value of the inverter reverse electromotive force generated by the bottom electric drive servo in the previous control cycle and convert it into a phase lag characterization parameter. Adjust the sliding time window width of the current control cycle according to the phase lag characterization parameter so that the sliding time window width is an integer value after rounding and converting the sum of the reference window constant and the phase lag characterization parameter. Step S103: The gradient of the tracking deviation vector is introduced into the sliding time window after the phase lag characterization parameter is adjusted to perform discrete integration calculation, and the transient cumulative energy scalar is generated. Step S104: Compare the transient cumulative energy scalar with the preset integral energy gating threshold. If the transient cumulative energy scalar crosses the integral energy gating threshold within 5 consecutive control cycles, issue a trajectory reconstruction trigger command to activate the path space reconstruction control program. Use the linear feedforward compensation matrix to convert the tracking deviation vector gradient into the attitude angle expectation correction increment matrix and inject it into the axis servo loop. Otherwise, switch to the normal route hold control mode and call the proportional-derivative control algorithm to maintain stable flight within the safe envelope.

[0021] Preferably, step S101 includes the following sub-steps: Step S1011, when the unmanned aerial vehicle encounters local electromagnetic field interference causing intermittent occlusion of spatial deviation data and discrete loss, the airborne inertial control state discrete prediction subroutine is started; Step S1012, the state transition matrix parameters are calculated based on the kinematic multi-order derivative vectors retained before the disconnection; Step S1013, the lost data sequence is filled in by the prediction output of the airborne inertial control state discrete prediction subroutine, and the tracking deviation vector gradient is determined.

[0022] Preferably, in step S104, the integral energy gating threshold is determined by the discrete integral energy criterion; the transient cumulative energy scalar obtained by discretizing and integrating the gradient of the continuously input tracking deviation vector within the sliding time window is used to reflect the high-frequency disturbance energy intensity of the control loop when the UAV is tracking at close range in a narrow inspection channel.

[0023] Preferably, in step S104, the path space reconstruction control program calls the preset adaptive attitude control loop toolbox; the tracking deviation vector gradient is input into the linear feedforward compensation matrix, and the desired attitude angle correction increment matrix is ​​calculated by mapping.

[0024] Preferably, the adaptive attitude control loop toolbox is equipped with a maximum switching frequency limit parameter; the maximum switching frequency limit parameter monotonically constrains the maximum action frequency based on the transient power supply capability of the airborne physical energy system; when the axis servo loop load fluctuates due to large-angle trajectory reconstruction, the system automatically reduces the maximum action switching frequency band of the attitude axis according to the power battery voltage drop trend, limits the overshoot amplitude of the electric drive servo motor under low voltage conditions, suppresses the mechanical resonance overshoot of the axis servo loop, and ensures the action coordination of the motion mechanism under limited airborne computing power and energy constraints.

[0025] Preferably, in step S104, issuing a trajectory reconstruction trigger command to activate the path space reconstruction control program includes the following sub-steps: Step S1041, determining the occurrence of an out-of-bounds discrete event through a discrete threshold arbitration state machine, and activating the path space reconstruction control program; Step S1042, substituting the tracking deviation vector gradient into the linear feedforward compensation matrix, and generating an attitude angle expected correction increment matrix through nonlinear mapping; Step S1043, injecting the attitude angle expected correction increment matrix into the axis servo loop in real time, driving the unmanned aerial vehicle to change its motion attitude to avoid the center of strong electromagnetic interference.

[0026] Preferably, in step S104, switching to the conventional flight path holding control mode includes the following sub-steps: Step S1044, when the transient cumulative energy scalar does not cross the integral energy gating threshold for 5 consecutive control cycles, the discrete threshold arbitration state machine suspends the subsequent spatial deformation matrix calculation; Step S1045, the system automatically switches to the conventional flight path holding control mode by reducing dimensions, calling the proportional-derivative control algorithm to maintain flight path tracking control within the safe envelope range, filtering electromagnetic flypoint noise caused by intermittent occlusion, and avoiding frequent switching of the discrete threshold arbitration state machine between different control commands.

[0027] Preferably, before the unmanned aerial vehicle takes off, the state space prediction operator and the discrete threshold arbitration state machine are initialized through the onboard embedded real-time operating system, a time-sensitive closed-loop support environment is established for the state space prediction operator and the discrete threshold arbitration state machine, and a safety envelope range is established according to the spatial geometric constraint distribution of the inspection channel.

[0028] Preferably, the tracking deviation vector gradient is calculated in real time based on the deviation between the current attitude angle of the UAV and the pre-stored reference inspection route; the deviation includes position deviation, velocity deviation and angle deviation.

[0029] Example 1: A path planning and optimization method for a UAV power line inspection task is applied to a flight control device carried by an airborne embedded real-time operating system. When the UAV conducts close-range inspections along high-voltage power transmission corridors and transmission towers, geometrical abrupt changes around the transmission towers will form irregular micro-airflow vortices. The strong electromagnetic field generated by the high-voltage conductors will cause random jitter noise and intermittent packet loss in the absolute space three-dimensional coordinate sequence collected by the airborne sensing system. The spatial deviation data will then exhibit non-steady-state multi-dimensional boundary crossing characteristics. When correcting flight deviations under the constraints of limited airborne computing power and airborne energy system, the flight control device receives the absolute space three-dimensional coordinate sequence in real time, uses the kinematic state space state matrix equation to remove high-frequency positioning jitter components, calculates the cleaned three-dimensional trajectory position data of the current time step, and determines it as the input parameter for the subsequent discrete reconstruction of the control loop.

[0030] When an unmanned aerial vehicle (UAV) passes close to a power transmission corridor and encounters a sudden, strong local electromagnetic field interference, causing discontinuous occlusion and discrete loss of the absolute space three-dimensional coordinate sequence, the system automatically switches to and activates the onboard inertial control state discrete prediction subroutine. This subroutine periodically extracts a set of cleaned three-dimensional trajectory position data from three consecutive control cycles remaining at the last moment before the disconnection. Using first- to third-order finite-difference time-domain algorithms, it sequentially discretizes and calculates the motion position, velocity, acceleration, and third-order jerk corresponding to the current time step, and combines them to construct a one-dimensional, fourth-order kinematic multi-order derivative vector. The determination rule for the state transition matrix parameters is defined as a four-dimensional square matrix. The parameter elements of this matrix are adaptively assembled based on the first-order classical kinematic Taylor series expansion equation, where the parameter elements in the first row of the matrix are respectively composed of constants. Item 1, along with the time-domain coefficients of the first, second, and third-order terms of the current single-step discrete control cycle time step, is used to characterize the inertial extrapolation of the position state dimension. The parameter elements in the second to fourth rows are set as the normal first-order physical evolution time-series coefficients of velocity, acceleration, and jerk. The processor substitutes the current fixed control cycle step into the coefficient structure of this matrix to complete the in-situ update and assembly of the values ​​of each product parameter in the matrix. Subsequently, the kinematic multi-order derivative vector is used as the column vector input and matrix multiplication is performed with the four-dimensional state transition matrix after the parameters are determined in the arithmetic logic unit to extrapolate the estimated value of the absolute spatial three-dimensional coordinates lost due to the signal discontinuity occlusion at the current time step. This predicted output is used to fill in the lost data sequence and restore the continuous time-domain difference, thereby determining the tracking deviation vector gradient.

[0031] The flight control unit uses a differential processing mechanism to calculate the difference between the cleaned 3D trajectory position data and the pre-stored baseline inspection route to determine the real-time deviation vector. Then, it calculates the tracking deviation vector gradient for the current control cycle by using the time-domain difference of the real-time deviation vectors from adjacent control cycles. To ensure a closed-loop feedback between the sampling bandwidth of the pre-discrete operator and the delay state of the electric drive servo, the flight control unit reads the discrete residual value of the inverter reverse electromotive force generated by the electric drive servo control mechanism in the previous adjacent control cycle in real time and converts it into a phase lag characterization parameter. Based on the phase lag characterization parameter Width of the sliding time window for the current period Adjust the sliding time window width. The following calculation formula is satisfied: ,in, The current control cycle number is The width of the sliding time window, which is an integer number of cycles; This is the rounding function; This is the preset reference window constant in the storage unit; This is a parameter used to characterize the nonlinear phase lag of the electric drive servo motor, which was detected and determined in the previous adjacent control cycle.

[0032] In practical operation, when the discrete residual value of the inverter reverse electromotive force at the bottom layer is converted into a dimensionless phase lag characterization parameter, the three-phase power inverter bridge inside the electric drive servo motor captures the discrete reverse residual current in the inverter bridge freewheeling diodes in real time through the phase current sampling resistor integrated in the hardware circuit under reverse bias conditions such as deceleration, braking, or high-gain high-frequency commutation. 50 microseconds before the end of each sampling control cycle, a hardware timer triggers the onboard analog-to-digital converter to acquire the absolute value of the discrete residual voltage of the induced inverter reverse electromotive force superimposed across the three-phase coil windings, in volts. Subsequently, the onboard microprocessor runs a normalization calculation program to convert the residual voltage into an absolute value. The logarithmic value is used as the numerator and divided by the preset rated operating voltage reference value of the electric drive servo system to obtain the dimensionless electromagnetic saturation coefficient, which reflects the degree of hysteresis of the actuator. The processor then calls the exponential physical mapping curve that has been pre-calibrated and stored in the flash memory, and substitutes the electromagnetic saturation coefficient as an independent variable into the exponential operation function to perform time-domain delay analysis calculation. The calculation output is used to characterize the physical hysteresis delay time caused by the increase of hardware damping in the current control loop. This value is normalized into a dimensionless parameter inside the calculation and output as a phase hysteresis characterization parameter, which is converted from the underlying electrical physical characteristics into the parameters of the high-level adaptive window width algorithm.

[0033] The flight control unit introduces the tracking deviation vector gradient of the current control cycle into a sliding time window adjusted by the phase lag characterization parameter, calculates the discrete integral in the time domain, and thus determines the transient cumulative energy scalar that reflects the energy intensity of the control loop disturbance when the UAV is tracking in the inspection channel. The discrete threshold arbitration state machine compares this transient cumulative energy scalar with the preset integral energy gating threshold. When the transient cumulative energy scalar is greater than the integral energy gating threshold for 5 consecutive control cycles, the discrete threshold arbitration state machine outputs a trajectory reconstruction trigger command to start the path space reconstruction control program. The tracking deviation vector gradient is converted into an attitude angle expectation correction increment matrix using a linear feedforward compensation matrix, and injected into the axis servo loop of the electric drive servo in real time to change the motion command output of the flight control loop.

[0034] Specifically, the nonlinear mapping transformation achieved through the linear feedforward compensation matrix employs a multi-model piecewise linearization approximation strategy to reconstruct the high-dimensional nonlinear spatial trajectory control flow. The matrix parameters in the linear feedforward compensation matrix are not fixed constants, but rather sets of conditional control coefficients pre-calibrated offline and stored in a non-volatile memory register lookup table. During online system operation, the central processing unit uses the attitude angular position, velocity, and angle multidimensional deviation vectors calculated at the current time step, along with the real-time wind speed of the external crosswind field, as nonlinear scheduling variables. These variables are input into a pre-defined hyperbolic tangent activation transformation function for nonlinear continuous addition. Weight calculation: Based on the calculated nonlinear decay distribution weights, the system dynamically retrieves and performs linear interpolation in a lookup table in real time, dynamically calculating the set of control matrix coefficients corresponding to the current control cycle. This completes the online dynamic assembly and real-time update of the linear feedforward compensation matrix parameter elements. Subsequently, the real-time tracking deviation vector gradient is multiplied by the dynamically updated feedforward compensation matrix using standard algebraic matrix multiplication. This low-computational-consumption linear matrix multiplication completes the high-dimensional nonlinear manifold space mapping from the input deviation gradient to the output attitude angle expected correction increment matrix, avoiding the inability of steady linear transformation to directly output nonlinear control response.

[0035] When the load fluctuates due to large-angle trajectory reconstruction in the axis servo loop, the maximum switching frequency limit parameter in the adaptive attitude control loop is limited by the current power battery output voltage envelope. The flight control device reduces the maximum action switching frequency band of the attitude axis according to the power battery voltage drop trend, and limits the overshoot action amplitude of the electric drive servo under low voltage conditions, suppresses the mechanical resonance overshoot of the axis servo loop, and weakens the dynamic causes of saturation oscillation in the control loop. When the transient cumulative energy scalar does not meet the judgment condition that it is greater than the integral energy gating threshold for 5 consecutive control cycles, the discrete threshold arbitration state machine suspends the subsequent path space reconstruction control program and switches to the normal route holding control mode. It calls the proportional-derivative control algorithm to maintain stable flight within the safe envelope range, and filters the electromagnetic interference noise caused by intermittent packet loss. This causes the UAV yaw error to stop diverging at 85ms after the path space reconstruction control program is started at 42ms, and converges to within 0.08m at 145ms. The control waveform shows a noisy asymmetric convergence state that conforms to the aerodynamic damping characteristics.

[0036] In actual operation, the discrete threshold arbitration state machine is a finite state control logic entity residing in the CPU core of the flight control device. It possesses mutually exclusive operating modes. Its internal topology consists of two mutually exclusive state registers: a conventional route-keeping control node and a path space reconstruction control node. Out-of-bounds discrete events are sudden quantized events reflecting high-frequency disturbance energy intensity in the control loop exceeding the safety envelope limit. The state transition rules and quantization trigger criteria for out-of-bounds discrete events of this discrete threshold arbitration state machine are as follows: Under normal conditions, the state machine is in the conventional route-keeping control node and the proportional-derivative control algorithm is activated. When the state machine compares the input transient cumulative energy scalar with the integral energy gating threshold fixed in the register in real time, if the transient cumulative energy... If the scalar value exceeds the integral energy gating threshold for five consecutive discrete control cycles, the software counter triggers an out-of-bounds overflow signal. Based on this, the system determines that an out-of-bounds discrete event has occurred. The discrete threshold arbitration state machine suspends the original conventional control flow, triggers a state transition in place, and forcibly jumps to the path space reconstruction control node. It outputs a trajectory reconstruction trigger command to activate the path space reconstruction control program and drives the electric drive servo to deflect to avoid the strong interference center. Conversely, in subsequent operation, once it is detected that the transient accumulated energy scalar has not crossed the integral energy gating threshold in any cycle, the counter is cleared, the state machine automatically triggers a reverse state transition, and returns to the conventional route holding control node. Through deterministic discrete flow control, high-frequency noise is filtered out, and high-frequency shuttle oscillations of the state are eliminated.

[0037] Example 2: In this example, the flight control device is still applied to the unmanned aerial vehicle carried by the airborne embedded real-time operating system. When the unmanned aerial vehicle maintains flight path tracking along the high-voltage power transmission corridor and power transmission tower at close range, the geometrical abrupt changes around the power transmission tower form irregular micro-airflow vortices. The strong electromagnetic field of the high-voltage conductor causes random jitter noise and intermittent packet loss in the absolute space three-dimensional coordinate sequence collected by the airborne sensing system. The flight control device receives the absolute space three-dimensional coordinate sequence in real time, uses the kinematic state space state matrix equation to remove high-frequency positioning jitter components, calculates the cleaned three-dimensional trajectory position data of the current time step, and uses it as the input parameter for subsequent discrete reconstruction of the control loop.

[0038] To determine the reference window constant The value range is determined by the numerical determination process of the central processing unit (CPU) within the flight control device. Constraints are set on the discrete sampling delay time of the control loop and the real-time floating-point data processing load of the CPU. Under conditions where the turbulence intensity of the irregular micro-airflow vortex generated in the narrow inspection channel is between 0.15 and 0.35, and the rate of change of energy output density in the radio frequency section of the strong electromagnetic field of the high-voltage conductor is not less than 15%, the CPU is tested incrementally within the discrete integer range of 2 to 35. The test results show that when When the set value is less than 5, the sliding time window width is too small, and high-frequency electromagnetic flying point noise is difficult to smooth out within the window; when When the set value is greater than 25, the sliding time window width is too large, and the accumulated phase lag of the control loop exceeds the damping critical point of the axis servo loop. Therefore, the reference window constant is adjusted. The working envelope range was determined to be 5 to 25, and in the optimization test under normal operating conditions, The discrete entity setting value is fixed at 10, which serves as the initial input source for subsequent experimental schemes.

[0039] To verify the effect of the discrete residual value of the inverter's back electromotive force on the sliding sampling bandwidth, the test process included a fixed sampling window control group, an out-of-range control group exceeding the working envelope value, and an experimental group containing multi-level interference gradients. Each test group ran for 300 seconds in the same hardware-in-the-loop simulation environment, continuously recording the dynamic discrete response data of each physical intermediate variable under a timing constraint of a 5ms sampling period. The electromagnetic interference intensity was 20dB. In the control group with a fixed sampling window set to 10, the phase lag characterization parameter The average measurement value is 0, and the sliding time window width is... The average sampling value remained at 10. Since this group did not introduce an adaptive adjustment mechanism for the phase lag characterization parameter, when the response lag of the electric drive servo increased due to the aerodynamic irregular vortex, the sliding time window width did not expand with the change of the servo's physical state. The transient cumulative energy scalar measurement value was 4.12. The high-frequency positioning jitter noise directly entered the threshold arbitration stage. The discrete threshold arbitration state machine frequently experienced erroneous arbitration switching between 55ms and 115ms. The control loop parameters were continuously in the high-gain fine-tuning phase, and the axis servo loop experienced deep saturation oscillation at 120ms. The convergence time of this group was 215ms, and the final yaw error remained at 0.24m.

[0040] When the electromagnetic interference intensity is the same at 20dB, In the out-of-range control group set to 2, the phase lag characterization parameter The average measurement value is 0.15, and the sliding time window width is... The average sampling value was 2. Since this setpoint was below the lower limit of the working envelope interval (5), the sliding time window width was insufficient to smooth high-frequency noise. The transient cumulative energy scalar measurement was 1.88, exhibiting random jumps. This scalar failed to exceed the integral energy gating threshold for five consecutive control cycles, and the path space reconstruction control program did not trigger in time when deviating from the critical point. The convergence time for this group was 195ms, and the final yaw error was 0.31m. This was achieved when the electromagnetic interference intensity was still 20dB. In another out-of-range control group set at 35, the phase lag characterization parameter The average measurement value is 0.14, and the sliding time window width is... The average sampled value increased to 40. Since this set value is higher than the upper limit of the working envelope interval of 25, the time domain delay generated by the sliding time window width calculation is superimposed on the electromechanical constant of the servo itself, and the transient cumulative energy scalar measurement value increased to 18.52. The injection phase lag of the attitude angle expected correction increment matrix exceeded the damping tolerance. The convergence time of this group was extended to 310ms, and the final yaw error expanded to 0.45m, showing overload control divergence.

[0041] In the experimental group, the baseline window constant When the boundary endpoints and normal median of the coverage working envelope are both 20 dB, In experimental group one, where the value is set to 5, the phase lag characterization parameter is... The average measurement value is 0.12, and the sliding time window width is... The average sampling value was 6, the transient cumulative energy scalar measurement value was 6.45, the convergence time was 158 ms, and the final yaw error was 0.09 m. In experimental group two, where the value was set to 10, the phase lag characterization parameter was... The average measurement value is 0.15, and the sliding time window width is... The average sampling value was 12, the transient cumulative energy scalar measurement value was 8.24, the convergence time was 145ms, and the final yaw error was 0.08m. In experimental group three, where the value was set to 15, the phase lag characterization parameter was... The average measurement value is 0.16, and the sliding time window width is... The average sampling value was 17, the transient cumulative energy scalar measurement value was 9.11, the convergence time was 141 ms, and the final yaw error was 0.07 m. In experimental group four, where the value was set to 25, the phase lag characterization parameter was... The average measurement value is 0.14, and the sliding time window width is... The average sampling value was 29, the transient cumulative energy scalar measurement value was 12.36, the convergence time was 162ms, and the final yaw error was 0.09m.

[0042] The above four sets of data show that when When the operating envelope ranges from 5 to 25, the system adjusts the sliding time window width based on the discrete residual value of the inverter back electromotive force. The adjustment is between 6 and 29, which smooths out the spatial high-frequency jitter components at the input end during the discrete integration process. At the same time, the physical state feedback of the electrically driven servo motor offsets the physical lag of the actuator induced by the large load action. The corresponding convergence time is maintained in the range of 141ms to 162ms, and the yaw error is ultimately kept within the safe envelope range of 0.07m to 0.09m.

[0043] In fixed With a set value of 10, the test further varied the intensity of the externally injected electromagnetic interference. When the electromagnetic interference intensity was 10 dB, the phase hysteresis characterization parameter... The average measurement rose to 0.28, and the sliding time window width... The average sampling value is 13, the transient cumulative energy scalar measurement value is 11.47, the convergence time is 152ms, the final yaw error is 0.08m, and the phase lag characterization parameter is 30dB. The average measurement value decreased to 0.05, and the sliding time window width... The average sampling value was 11, the transient cumulative energy scalar measurement value was 5.88, the convergence time was 138ms, the final yaw error was 0.06m, and the electromagnetic interference intensity was 20dB. The data from test group two, set to 10, show that as the intensity of externally injected electromagnetic interference changes from the low-noise state corresponding to 30dB to the high-noise state corresponding to 10dB, the phase lag characterization parameter... The sliding time window width increases from 0.05 to 0.28. The average sample value was widened from 11 to 13, and the transient cumulative energy scalar remained in the range of 5.88 to 11.47.

[0044] As can be seen from the above test process, after adopting the method of this embodiment, the discrete residual value of the inverter back electromotive force, as the closed-loop driving source for sliding sampling bandwidth reconstruction, can adjust the width of the sliding time window according to the physical hysteresis state of the electric drive servo. Compared with the control group with a fixed sampling window, the experimental group reduced the final yaw error from 0.24m to 0.07m to 0.09m under the same 20dB electromagnetic interference intensity; compared with the out-of-range control group, When the operating envelope range of 5 to 25 is reached, the control loop does not exhibit delayed triggering due to an excessively narrow window, nor does it exhibit overload control divergence due to an excessively wide window. This data indicates that in a complex variable flow field environment with strong interference and narrow channels, the flight control device can adjust the sliding time window width according to the state of the electric drive servo motor, so that the spatial motion state of the unmanned aerial vehicle remains converged and the tracking accuracy of the flight path is maintained.

[0045] Example 3: Before the first flight of the unmanned aerial vehicle (UAV), the flight control device runs the integral energy gating threshold calibration process. During calibration, the UAV remains stationary on the ground, and the surrounding high-voltage power lines apply radio frequency radiation normally. The onboard microprocessor and the absolute space three-dimensional positioning sensor constitute the hardware basis for the calibration test. The onboard microprocessor has a main frequency of 1.2 GHz and integrates a floating-point arithmetic unit. The absolute space three-dimensional positioning sensor has a maximum discrete sampling frequency of 200 Hz and a physical measurement resolution of 0.01 m. Under stationary calibration conditions, the onboard microprocessor collects the absolute space three-dimensional coordinate sequence output by the absolute space three-dimensional positioning sensor within 1000 consecutive sampling periods and uses it as the original positioning noise sequence. Subsequently, the standard deviation of each spatial axial component in the sequence is calculated using the time-domain statistical variance calculation method. The maximum value among the standard deviation components is determined as the eigenvalue of the basic electromagnetic noise disturbance. ,in, The eigenvalues ​​are the electromagnetic noise disturbance eigenvalues ​​of the absolute spatial positioning sensor.

[0046] When the unmanned aerial vehicle (UAV) is controlled to fly along a pre-stored reference inspection route, the onboard microprocessor collects the current axial component deviation through the absolute space three-dimensional positioning sensor at a discrete sampling period of 5ms. It calculates the component difference between the cleaned three-dimensional trajectory position data and the pre-stored reference inspection route to determine the real-time deviation vector sequence. The flight control device introduces this real-time deviation vector sequence into a basic sliding time window with a time width of 10 cycles to calculate the discrete integral in the time domain. From this, the normal cumulative energy change sequence characterizing the impact of normal aerodynamic disturbances is obtained. Then, the maximum value of this sequence within a 300s measurement period is extracted, and this maximum value is determined as the upper limit of the dynamic airflow disturbance tolerance. .

[0047] After completing the aforementioned static calibration and simulated inspection conditions, the flight control unit, according to the discrete integral energy criterion, sets the eigenvalues ​​of the basic electromagnetic noise disturbance. With dynamic airflow disturbance tolerance upper limit Sum the results and then multiply by the preset safety redundancy adjustment factor. Determine the integral energy gating threshold. Integral energy gating threshold Satisfy the following formula: ,in, The integral energy gating threshold is... The preset safety redundancy adjustment factor is set to 1.25. The eigenvalues ​​of electromagnetic noise disturbance of the absolute spatial positioning sensor, determined under previous static conditions. This is the upper limit of the dynamic airflow disturbance tolerance calculated from the simulated inspection conditions.

[0048] The flight control unit will set the integral energy gating threshold. The energy scalar value of the transient accumulated energy in the current cycle remains at 4.85 when the UAV passes close to a sudden irregular vortex interference zone with a turbulence intensity of 0.35 around a power transmission tower, stored in the non-volatile memory register of the discrete threshold arbitration state machine. This value does not exceed the integral energy threshold. The quantitative judgment boundary is determined, the discrete threshold arbitration state machine remains in an untriggered state, and the trajectory reconstruction trigger command is suspended. As a result, the high-frequency random jitter noise of the sensing system does not trigger the control loop to switch erroneously, and the second-order time delay commutation loss conversion value of the electric drive servo control mechanism is reduced accordingly. Under the crosswind condition of 12m / s, the pitch angle overshoot fluctuation of the UAV is controlled within 1.5°, and the spatial yaw error of the flight path tracking is stabilized within 0.08m.

[0049] Example 4: This example combines Figures 1 to 2 This paper describes a path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks, such as... Figure 1As shown, first, step S101 is executed to acquire the absolute space three-dimensional coordinate sequence of the unmanned aerial vehicle, calculate the real-time deviation vector between the absolute space three-dimensional coordinate sequence and the pre-stored reference inspection route, and calculate the tracking deviation vector gradient of the current control cycle by the time-domain difference of the real-time deviation vector of adjacent control cycles; then, it proceeds to step S102 to detect the discrete residual value of the inverter reverse electromotive force generated by the underlying electric drive servo in the previous control cycle and convert it into a phase lag characterization parameter, and adjust the sliding time window width of the current control cycle according to the phase lag characterization parameter, so that the sliding time window width is an integer value obtained by rounding down the sum of the reference window constant and the phase lag characterization parameter; then, it proceeds to... Step S103: The tracking deviation vector gradient is introduced into the sliding time window after adjustment of the phase lag characterization parameter to perform discrete integral calculation, producing a transient cumulative energy scalar; finally, it flows to step S104, where the transient cumulative energy scalar is compared with the preset integral energy gate threshold. If the transient cumulative energy scalar crosses the integral energy gate threshold for 5 consecutive control cycles, a trajectory reconstruction trigger command is issued to activate the path space reconstruction control program. The tracking deviation vector gradient is converted into an attitude angle expectation correction increment matrix using a linear feedforward compensation matrix and injected into the axis servo loop. Otherwise, it switches to the conventional route hold control mode and calls the proportional-derivative control algorithm to maintain stable flight within the safe envelope range.

[0050] like Figure 2 As shown, the system upon which this method relies includes an airborne sensing system, a flight control device, an airborne physical energy system, and underlying actuators. The airborne sensing system includes an absolute space three-dimensional positioning sensor and a three-axis microelectromechanical system accelerometer. The flight control device includes an airborne embedded real-time operating system, an airborne microprocessor, and non-volatile memory. The airborne physical energy system includes a power battery. The underlying actuators include an electric drive servo control mechanism and a three-phase power inverter bridge. In the data interaction and physical flow relationship between the components, the absolute space three-dimensional positioning sensor outputs an absolute space three-dimensional coordinate sequence to the airborne microprocessor, the power battery outputs the power battery voltage to the airborne microprocessor, the airborne microprocessor outputs the attitude angle expected correction increment matrix to the electric drive servo control mechanism to control the underlying actuators, and the three-phase power inverter bridge in the underlying actuators feeds back the discrete residual value of the inverter reverse electromotive force to the airborne microprocessor. The data calculation and status monitoring of the entire process are coordinated and completed by the airborne microprocessor with the support of the airborne embedded real-time operating system and the storage support of the non-volatile memory.

[0051] Example 5: When the unmanned aerial vehicle (UAV) is in the pre-launch calibration state before unlocking, the flight control device runs an offline zero-bias calibration program to eliminate non-axial installation errors of the sensors. The program's raw data input receives the discrete gravity acceleration vector sequence output by the triaxial MEMS accelerometer in a stationary state. The triaxial MEMS accelerometer is installed on a calibration pedestal with a levelness variation of less than 0.01°. The onboard central processing unit within the flight control device performs a moving average filtering smoothing calculation on the discrete gravity acceleration vector sequence for 500 consecutive control cycles to obtain the static installation bias angle matrix between the sensor coordinate system and the body axis system. ,in, The static installation bias angle matrix is ​​written in-situ into the zero bias configuration register of the non-volatile memory as a calibration parameter, and is used as a spatial rotation alignment reference value for real-time bias vector transformation in subsequent flight calculations.

[0052] When the unmanned aerial vehicle (UAV) tracks the high-voltage power transmission corridor along its inspection route, the sensors experience zero-point thermal drift due to ambient temperature changes. The onboard central processing unit (CPU) initiates a dynamic zero-baseline adjustment program, updating the baseline values ​​in the zero-bias configuration register. Within a steady-state flight window where the UAV maintains uniform straight-line flight for more than 3 seconds and the tracking deviation vector gradient is less than 0.02 m / s, the system periodically collects the median component of the discrete residual values ​​of the inverter reverse electromotive force output by the electric drive servo control mechanism in the previous adjacent control cycle, calculating the temperature drift correction increment matrix in the zero-bias configuration register. Subsequently, the static installation deflection matrix is... The incremental matrix of temperature drift correction is combined with the spatial axial correction command, which is then used to drive the control word register of the electric drive servo control mechanism in place to complete the hardware zero-position offset, thereby offsetting the control deviation caused by structural thermal deformation. This ensures that the residual yaw error of the spatial trajectory after 300s of continuous inspection is stably controlled within 0.05m.

[0053] In actual operation, the electric drive servo control mechanism undergoes high-frequency commutation under multi-field coupled alternating loads such as sudden unsteady flow fields. The internal commutation heat generation increases accordingly. The median component of the discrete residual value of the inverter reverse electromotive force is monotonically positively correlated with the electromechanical cumulative heat loss of the electric drive servo. The local heat generated by this heat loss is conducted upward through the carbon fiber composite skeleton of the airframe and the aluminum alloy servo mounting bracket, and then to the rigid calibration platform of the front-end sensing system. This causes the UAV skeleton structure to undergo micron-level physical thermal expansion deformation. At the same time, the positioning sensor chips such as the accelerometer of the nearby rigidly mounted triaxial microelectromechanical system are heated unevenly, forming a small local temperature gradient. This causes the original positioning coordinates output by the sensitive elements to undergo zero-point thermal drift.

[0054] The flight control device is based on a quasi-steady-state equilibrium physical model of solid thermal conduction. It uses the median component of the discrete residual value of the inverter reverse electromotive force collected in real time as an indirect characterization of the intensity of local heat sources inside the structure. It performs time-domain smoothing filtering calculations by using a preset time constant weight matrix that decays with the thermal conduction delay of spatial distance. It solves the minute deformation angle accumulated on the spatial geometric axis of the positioning sensor due to heat conduction of the local structure, as well as the zero bias voltage increment of the positioning sensor's sensitive chip due to temperature rise. The aforementioned minute deformation angle and zero bias voltage increment together constitute the temperature drift correction increment matrix, which enables the electromagnetic heating state feedback of the underlying physical actuator to be used for compensation of the physical environment temperature drift and mechanical thermal deformation deviation of the front-end sensor, and completes hardware zero-position offsetting.

[0055] Example 6: The current flight control device operates during long-endurance power inspection missions of unmanned aerial vehicles (UAVs) and its parameter recalibration process. This process corrects the accumulated impedance offset caused by long-term alternating airflow oscillations and the commutation heating of the electrically driven servo motors. When the UAV continuously tracks the flight path, the impedance state of the electrically driven servo motors changes with operating time and thermal load, causing the phase lag characterization parameters detected and determined in the previous adjacent control cycle to... A continuous bias is generated; if this bias directly enters the sliding time window width adjustment process, the position control loop is prone to tracking divergence.

[0056] To ensure uninterrupted flight missions and maintain continuous anti-jamming tracking, the onboard processor automatically reads the phase lag characterization parameters from the past 2000 control cycles when the UAV has been continuously tracking the flight path for 5000 seconds. The arithmetic mean is calculated in memory to filter out transient fluctuation components. Subsequently, a scaling factor for offsetting impedance aging is calculated based on the decimal deviation of this arithmetic mean. Then use the scaling factor For the initial reference window constant Perform product correction to obtain the dynamic reference window constant after online calibration. Dynamic reference window constant The following calculation formula is satisfied: ,in, This is the dynamic reference window constant after online calibration. This is the rounding function; The initial reference window constant is preset in the storage cell, and its set value is 10; This is the dimensionless scaling factor derived from the arithmetic mean.

[0057] When verifying the stability of spatial motion control under different hardware aging and thermal drift conditions through parameter recalibration, the hardware-in-the-loop simulation platform conducted long-period multi-level gradient perturbation tests by adjusting the impedance offset simulation parameter at the input. When the simulated impedance offset parameter monotonically increased from 0.02 in the initial unbiased state to 0.32 in the severely aged state, the scaling factor calculated by the processor... The dynamic baseline window constant decreases monotonically from 1.00 to 0.72. The correction cutoff impedance bias from 10 to 7 affects the sliding time window width. The resulting time delay amplification path keeps the probability of false triggering switching of the path space reconstruction control program in the sudden vortex flow field below 1%; under the undervoltage condition where the remaining output power voltage envelope of the airborne power battery decays from 24.2V to 19.8V due to deep discharge, the overshoot response damping ratio of the axis servo circuit when completing large-angle route correction is still maintained in the range of 0.75 to 0.82, and the yaw error of the UAV in spatial three-dimensional trajectory tracking at the end of the 5000s inspection is controlled within 0.06m.

[0058] To address the frequency limiting control under the declining trend of power battery voltage, the airborne central processing unit reads the real-time output voltage value of the power battery bus collected and reported by the hardware power management module in real time with a fixed interrupt cycle of 20ms. When the real-time voltage value is within the normal range of full charge rated voltage of 24.2V to 22.0V, the maximum switching frequency limit parameter is kept in place at the preset default maximum coefficient, the highest action switching frequency band of the attitude axis is locked at the full width of 50Hz, and the axis servo loop performs the fastest action response switching.

[0059] When the path space reconstruction control program is activated due to large-angle trajectory reconstruction, and the high current draw of the electric drive servo causes the real-time bus voltage to drop below 22.0V and continue to slide towards the low voltage of 19.8V, the central processing unit starts the linear derating limiting program. This program uses the difference between the current real-time voltage and the normal lower limit voltage of 22.0V as the input variable. According to the proportional attenuation rule that the highest action switching frequency band is monotonically reduced by 1.5Hz for every 0.1V decrease in bus voltage, the program dynamically updates the maximum switching frequency limit parameter and sends it offline to the axis servo circuit to perform derating frequency limiting control. When the real-time voltage further drops below or equals the 19.8V limit undervoltage judgment boundary, the frequency limiting control enters the hard cutoff protection mode. The highest action switching frequency band of the attitude axis is forcibly locked at the minimum safe holding frequency of 17Hz, thereby limiting the overshoot amplitude of the electric drive servo under low voltage conditions, avoiding mechanical resonance in the axis servo circuit, and ensuring the coordination of the control system under energy-constrained conditions.

[0060] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks, characterized in that, Includes the following steps: Step S101: Collect the absolute space three-dimensional coordinate sequence of the unmanned aerial vehicle, calculate the real-time deviation vector between the absolute space three-dimensional coordinate sequence and the pre-stored benchmark inspection route, and calculate the tracking deviation vector gradient of the current control cycle by the time domain difference of the real-time deviation vector of adjacent control cycles. Step S102: Detect the discrete residual value of the inverter reverse electromotive force generated by the bottom electric drive servo in the previous control cycle and convert it into a phase lag characterization parameter. Adjust the sliding time window width of the current control cycle according to the phase lag characterization parameter so that the sliding time window width is an integer value after rounding and converting the sum of the reference window constant and the phase lag characterization parameter. Step S103: The gradient of the tracking deviation vector is introduced into the sliding time window after the phase lag characterization parameter is adjusted to perform discrete integration calculation, and the transient cumulative energy scalar is generated. Step S104: Compare the transient cumulative energy scalar with the preset integral energy gating threshold. If the transient cumulative energy scalar crosses the integral energy gating threshold within 5 consecutive control cycles, issue a trajectory reconstruction trigger command to activate the path space reconstruction control program. Use the linear feedforward compensation matrix to convert the tracking deviation vector gradient into the attitude angle expectation correction increment matrix and inject it into the axis servo loop. Otherwise, switch to the normal route hold control mode and call the proportional-derivative control algorithm to maintain stable flight within the safe envelope.

2. The path planning and optimization method for UAV power line inspection tasks according to claim 1, characterized in that, Step S101 includes the following sub-steps: Step S1011, when the UAV encounters local electromagnetic field interference causing intermittent occlusion and discrete loss of spatial deviation data, the airborne inertial control state discrete prediction subroutine is started; Step S1012, the state transition matrix parameters are calculated based on the kinematic multi-order derivative vectors retained before the disconnection; Step S1013, the lost data sequence is filled in by the prediction output of the airborne inertial control state discrete prediction subroutine, and the tracking deviation vector gradient is determined.

3. The path planning and optimization method for UAV power line inspection tasks according to claim 1, characterized in that, In step S104, the integral energy gating threshold is determined by the discrete integral energy criterion; the transient cumulative energy scalar obtained by discretizing and integrating the gradient of the continuously input tracking deviation vector within the sliding time window is used to reflect the high-frequency disturbance energy intensity of the control loop when the UAV is tracking at close range in a narrow inspection channel.

4. The path planning and optimization method for UAV power line inspection tasks according to claim 1, characterized in that, In step S104, the path space reconstruction control program calls the preset adaptive attitude control loop toolbox; the tracking deviation vector gradient is input into the linear feedforward compensation matrix, and the desired attitude angle correction increment matrix is ​​calculated by mapping.

5. The path planning and optimization method for UAV power line inspection tasks according to claim 4, characterized in that, The adaptive attitude control loop toolbox is equipped with a maximum switching frequency limit parameter. The maximum switching frequency limit parameter monotonically constrains the maximum action frequency based on the transient power supply capability of the airborne physical energy system. When the axis servo loop load fluctuates due to large-angle trajectory reconstruction, the system automatically reduces the maximum action switching frequency band of the attitude axis according to the power battery voltage drop trend, limits the overshoot amplitude of the electric drive servo motor under low voltage conditions, suppresses the mechanical resonance overshoot of the axis servo loop, and ensures the motion coordination of the motion mechanism under limited airborne computing power and energy constraints.

6. The path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks according to claim 1, characterized in that, In step S104, issuing a trajectory reconstruction trigger command to activate the path space reconstruction control program includes the following sub-steps: Step S1041, determining the occurrence of an out-of-bounds discrete event through a discrete threshold arbitration state machine, and activating the path space reconstruction control program; Step S1042, substituting the tracking deviation vector gradient into the linear feedforward compensation matrix, and generating an attitude angle expected correction increment matrix through nonlinear mapping; Step S1043, injecting the attitude angle expected correction increment matrix into the axis servo loop in real time, driving the unmanned aerial vehicle to change its motion attitude to avoid the center of strong electromagnetic interference.

7. The method for path planning and optimization of UAV power line inspection tasks according to claim 1, characterized in that, In step S104, switching to the conventional flight path holding control mode includes the following sub-steps: Step S1044, when the transient cumulative energy scalar does not cross the integral energy gating threshold for 5 consecutive control cycles, the discrete threshold arbitration state machine suspends the subsequent spatial deformation matrix calculation; Step S1045, the system automatically switches to the conventional flight path holding control mode by reducing the dimension, calling the proportional-derivative control algorithm to maintain flight path tracking control within the safe envelope range, filtering electromagnetic flypoint noise caused by intermittent occlusion, and avoiding frequent switching of the discrete threshold arbitration state machine between different control commands.

8. The path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks according to claim 1, characterized in that, Before the unmanned aerial vehicle takes off, the state space prediction operator and discrete threshold arbitration state machine are initialized through the airborne embedded real-time operating system. A time-sensitive closed-loop support environment is established for the state space prediction operator and discrete threshold arbitration state machine, and a safety envelope range is established based on the spatial geometric constraint distribution of the inspection channel.

9. The path planning and optimization method for unmanned aerial vehicle (UAV) power line inspection tasks according to claim 1, characterized in that, The tracking deviation vector gradient is calculated in real time based on the deviation between the current attitude angle of the UAV and the pre-stored reference inspection route; the deviation includes position deviation, velocity deviation and angle deviation.

Images