A crossing unmanned aerial vehicle system with offline anti-interference and dynamic intelligent tracking

Abstract

This invention relates to the field of unmanned aerial vehicle (UAV) control technology and discloses a flight-ready UAV system with offline anti-interference and dynamic intelligent tracking. The system includes a data sensing module, a drive adjustment module, and a control processing module. The control processing module calculates the actual acceleration limit threshold of the aircraft and compares the observed acceleration parameter with this limit threshold. When the observed acceleration parameter exceeds the physical limit, feedback distortion is determined. By setting the feedback weight to 0, the feedback channel is disconnected, and the error integral term value of the adjustment loop is simultaneously locked to suppress integral saturation. During the feedback disconnection period, attitude angular rate data is used as a feedforward compensation reference to maintain stable drive commands. This invention avoids the risk of attitude collapse caused by transient interference, eliminates the accumulation of ill-conditioned errors, and achieves a seamless control transition between online adjustment and offline tracking.

Description

Technical Field

[0001] This invention relates to a drone system with offline anti-interference and dynamic intelligent tracking capabilities, belonging to the field of drone control technology. Background Technology

[0002] Current aerial drones typically employ cascaded proportional-integral-derivative (PID) control algorithms to achieve closed-loop control of attitude and trajectory. They calculate the motor control duty cycle by real-time fusion of feedback from the inertial measurement unit (IMU) and satellite navigation system to maintain stable operation in three-dimensional space. However, under complex conditions such as strong electromagnetic interference or sudden visual obstruction, the sensing unit generates erroneous measurements with drastic amplitude fluctuations and high-frequency oscillations. Existing control loops lack real-time verification of the physical validity of the feedback, causing the control unit to saturate in response to these non-physical deviations. This induces overload and aerodynamic stall in the actuator before offline task logic is invoked. This lack of physical immunity to transient high-frequency signal distortion is the fundamental reason for the low survivability of existing systems in extreme environments.

[0003] Schemes that involve adding multi-stage filters or redundant sensors often disrupt the system's dynamic equilibrium window due to the introduction of additional signal processing delays. Furthermore, adding shielding structures or high-performance hardware increases takeoff mass and weakens the aircraft's maneuverability, making it difficult for the system to achieve true physical steady state under stringent mass and power consumption constraints. In addition to hardware-level architectural limitations, the control system's algorithm logic has shortcomings in anti-interference robustness. For example, Chinese invention patent application CN113821059A discloses a fault-tolerant flight control system and method for multi-rotor UAVs, which relies on a primary / backup IMU and a multi-source sensor voting mechanism and residual verification to achieve fault diagnosis. Such schemes have underlying principle defects: the discrimination logic is deeply coupled with the relative consistency between sensors. When faced with common-mode interference scenarios caused by strong electromagnetic pulses, the sensing units synchronously generate non-physical deviations, causing the voting or residual verification mechanism to fail at the reference source level. Due to the lack of absolute physical constraints anchored to the inherent kinematic boundaries of the airframe, feedback distortion transients still induce the regulation loop response.

[0004] Therefore, how to construct a dynamic truncation and control takeover mechanism for feedback signals based on physical kinematic boundaries, so as to achieve the defense of the control loop against non-physical distortions and the maintenance of steady state under extreme interference conditions, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A drone system with offline anti-interference and dynamic intelligent tracking capabilities, comprising a data sensing module, a drive adjustment module, and a control processing module:

[0006] The control processing module is connected to the data sensing module and the drive adjustment module via signals, respectively.

[0007] The control processing module is used to obtain the total mass parameters of the aircraft and the maximum thrust-to-weight ratio parameters of the power system, and to calculate the actual acceleration limit threshold that the aircraft can reach when moving in three-dimensional space.

[0008] The control processing module is used to receive the spatial position feedback and velocity feedback output by the data sensing module, calculate the second-order finite difference value of the spatial position feedback within the current sampling period, or calculate the first-order finite difference value of the velocity feedback, and determine the observation acceleration parameter that characterizes the actual observation state.

[0009] The control processing module is used to implement the front-end legality verification of the adjustment loop. When the observed acceleration parameter is greater than the true acceleration limit threshold for two consecutive sampling periods, it determines that the feedback path has a non-physical distortion. The weight coefficients of the spatial position feedback and velocity feedback in the fusion adjustment algorithm are set to 0 to disconnect the feedback channel. At the same time, the error integral term in the attitude adjustment logic is locked to the historical calculation value of the sampling period before the interference judgment to suppress the integral saturation phenomenon caused by feedback distortion.

[0010] The control processing module is used to retrieve the attitude angular rate data output by the built-in inertial measurement unit during the period when the feedback channel is disconnected, and use it as the feedforward compensation reference input to the attitude adjustment logic to maintain the smooth duty cycle command output of the drive adjustment module until the newly calculated observed acceleration parameter falls back to within the true acceleration limit threshold, and then the closed-loop feedback adjustment of the feedback channel is restored.

[0011] Preferably, the control processing module is also used to keep the corresponding error integral register value constant during the period of locking the error integral term value, so as to block the accumulation of false deviations caused by feedback signal distortion, so that the duty cycle command output by the control processing module is kept within the physical safety envelope determined by the maximum thrust-weight ratio parameter, ensuring that the aircraft maintains its original aerodynamic attitude during the offline mission logic takeover transient, and avoiding the rotor overshoot surge caused by control signal overload.

[0012] Preferably, the data perception module includes a satellite positioning unit and a visual measurement unit; the control processing module is used to determine the dynamic intelligent tracking path of the UAV relative to the target through a multi-source data fusion algorithm based on the global coordinate data provided by the satellite positioning unit and the target feature vector provided by the visual measurement unit; the control processing module verifies the input parameters entering the multi-source data fusion algorithm through the real acceleration limit threshold in each adjustment cycle.

[0013] Preferably, the drive adjustment module includes a speed control unit and a power unit; the duty cycle command output by the control processing module is used to drive the speed control unit; when the feedback channel is disconnected, the control processing module controls the speed control unit by adjusting the feedforward compensation reference to maintain the stable speed of the power unit.

[0014] Preferably, the control processing module is also used to extract multimodal features of the image acquired by the visual measurement unit and calculate the target's trajectory within the next 100ms according to the trajectory prediction logic; when the feedback channel is disconnected, the control processing module updates the UAV's offline flight trajectory in real time according to the trajectory, so that the aircraft maintains its pointing towards the target in the absence of feedback.

[0015] Preferably, the control processing module is also used to monitor the signal quality parameters of the data sensing module; when the signal quality parameters are below 5dB and the duration exceeds 50ms, the control processing module switches to offline cruise mode and generates control signals for the drive adjustment module by reading the mission waypoint data in the built-in storage module and combining it with the attitude angular rate data.

[0016] Preferably, the control processing module stores a feedforward compensation mapping table, which records the basic duty cycle data corresponding to different flight speed ranges. During the period when the feedback channel is disconnected, the control processing module retrieves the corresponding basic duty cycle data according to the current flight speed and superimposes the deviation correction amount generated by the attitude angular rate data.

[0017] Preferably, the system also includes a communication link monitoring module, which is used to detect the continuity of external command signals; when the feedback channel is disconnected and the external command signal is interrupted, the control processing module activates the offline autonomous logic and uses the steady-state motion parameters before the feedback channel was disconnected as initial values ​​to calculate the flight trajectory.

[0018] Preferably, the system also includes an interference suppression and isolation module. The data sensing module is deployed at the center of the aircraft through the interference suppression and isolation module to attenuate the interference of mechanical vibration generated by the drive adjustment module on the acquisition of spatial position feedback.

[0019] Compared with the prior art, the beneficial effects of the present invention are:

[0020] 1. In dynamic intelligent tracking drones, by constructing a feedback cutoff mechanism based on physical kinematic boundaries at the front end of the control loop, the risk of attitude collapse caused by transient interference can be effectively avoided. In complex electromagnetic environments or visual obstruction conditions, the navigation and vision sensors generate false state variables with drastic amplitude changes and high-frequency oscillations before failure. The control unit extracts the second finite difference value of the feedback quantity and compares it with the actual acceleration limit threshold of the aircraft in real time. This allows the system to directly cut off the pathological feedback path at the bottom layer before the control command is converted into the overshoot signal of the power system, thereby eliminating the transient saturation of the actuator driven by the distorted variable and maintaining the aerodynamic stability of the aircraft during mode switching intervals.

[0021] 2. An error integral term freezing strategy is adopted in the attitude control algorithm to solve the ill-conditioned accumulation problem of the control loop under strong interference environment. By forcibly locking the integral term value in the cascade PID control logic to the steady-state historical calculation value before the interference occurs during the sampling period when the judgment feedback quantity is non-physically distorted, the path of high-frequency random noise entering the error accumulation link is blocked, and the irreversible integral saturation of the system due to false deviation in response is avoided. This ensures that the control output of the UAV always stays within the physical safety envelope during the process of switching from online closed-loop regulation to offline autonomous mission logic.

[0022] 3. By using a feedforward takeover mechanism based on high-frequency inertial measurement data, a seamless control transition between online closed-loop regulation and offline autonomous tracking is achieved. During the period when the feedback loop is truncated due to physical boundary verification, the control unit uses the angular rate data output by the built-in inertial measurement unit as the feedforward compensation reference input to the attitude adjustment loop. This maintains the smooth duty cycle output of the electronic speed controller in an open-loop or semi-closed-loop manner until the feedback observation value falls back to the physically reasonable range. This short-term takeover method based on inertial feedforward constructs a stable physical initial state for the final takeover of the offline autonomous mode, improving the system's survivability under extreme disturbances. Attached Figure Description

[0023] Figure 1 This is the control flowchart for physical boundary verification and feedforward compensation in this invention;

[0024] Figure 2 This is a system hardware architecture diagram of the interference suppression function of the present invention.

[0025] 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

[0026] 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.

[0027] A drone system with offline anti-interference and dynamic intelligent tracking capabilities includes a data sensing module, a drive adjustment module, and a control processing module.

[0028] The control processing module is connected to the data sensing module and the drive adjustment module via signals, respectively.

[0029] The control processing module is used to obtain the total mass parameters of the aircraft and the maximum thrust-to-weight ratio parameters of the power system, and to calculate the actual acceleration limit threshold that the aircraft can reach when moving in three-dimensional space.

[0030] The control processing module is used to receive the spatial position feedback and velocity feedback output by the data sensing module, calculate the second-order finite difference value of the spatial position feedback within the current sampling period, or calculate the first-order finite difference value of the velocity feedback, and determine the observation acceleration parameter that characterizes the actual observation state.

[0031] The control processing module is used to implement the front-end legality verification of the adjustment loop. When the observed acceleration parameter is greater than the true acceleration limit threshold for two consecutive sampling periods, it determines that the feedback path has a non-physical distortion. The weight coefficients of the spatial position feedback and velocity feedback in the fusion adjustment algorithm are set to 0 to disconnect the feedback channel. At the same time, the error integral term in the attitude adjustment logic is locked to the historical calculation value of the sampling period before the interference judgment to suppress the integral saturation phenomenon caused by feedback distortion.

[0032] The control processing module is used to retrieve the attitude angular rate data output by the built-in inertial measurement unit during the period when the feedback channel is disconnected, and use it as the feedforward compensation reference input to the attitude adjustment logic to maintain the smooth duty cycle command output of the drive adjustment module until the newly calculated observed acceleration parameter falls back to within the true acceleration limit threshold, and then the closed-loop feedback adjustment of the feedback channel is restored.

[0033] Preferably, the control processing module is also used to keep the corresponding error integral register value constant during the period of locking the error integral term value, so as to block the accumulation of false deviations caused by feedback signal distortion, so that the duty cycle command output by the control processing module is kept within the physical safety envelope determined by the maximum thrust-weight ratio parameter, ensuring that the aircraft maintains its original aerodynamic attitude during the offline mission logic takeover transient, and avoiding the rotor overshoot surge caused by control signal overload.

[0034] Preferably, the data perception module includes a satellite positioning unit and a visual measurement unit; the control processing module is used to determine the dynamic intelligent tracking path of the UAV relative to the target through a multi-source data fusion algorithm based on the global coordinate data provided by the satellite positioning unit and the target feature vector provided by the visual measurement unit; the control processing module verifies the input parameters entering the multi-source data fusion algorithm through the real acceleration limit threshold in each adjustment cycle.

[0035] Preferably, the drive adjustment module includes a speed control unit and a power unit; the duty cycle command output by the control processing module is used to drive the speed control unit; when the feedback channel is disconnected, the control processing module controls the speed control unit by adjusting the feedforward compensation reference to maintain the stable speed of the power unit.

[0036] Preferably, the control processing module is also used to extract multimodal features of the image acquired by the visual measurement unit and calculate the target's trajectory within the next 100ms according to the trajectory prediction logic; when the feedback channel is disconnected, the control processing module updates the UAV's offline flight trajectory in real time according to the trajectory, so that the aircraft maintains its pointing towards the target in the absence of feedback.

[0037] Preferably, the control processing module is also used to monitor the signal quality parameters of the data sensing module; when the signal quality parameters are below 5dB and the duration exceeds 50ms, the control processing module switches to offline cruise mode and generates control signals for the drive adjustment module by reading the mission waypoint data in the built-in storage module and combining it with the attitude angular rate data.

[0038] Preferably, the control processing module stores a feedforward compensation mapping table, which records the basic duty cycle data corresponding to different flight speed ranges. During the period when the feedback channel is disconnected, the control processing module retrieves the corresponding basic duty cycle data according to the current flight speed and superimposes the deviation correction amount generated by the attitude angular rate data.

[0039] Preferably, the system also includes a communication link monitoring module, which is used to detect the continuity of external command signals; when the feedback channel is disconnected and the external command signal is interrupted, the control processing module activates the offline autonomous logic and uses the steady-state motion parameters before the feedback channel was disconnected as initial values ​​to calculate the flight trajectory.

[0040] Preferably, the system also includes an interference suppression and isolation module. The data sensing module is deployed at the center of the aircraft through the interference suppression and isolation module to attenuate the interference of mechanical vibration generated by the drive adjustment module on the acquisition of spatial position feedback.

[0041] Example 1: In a scenario where a drone is performing a power line inspection mission, when the system faces high-intensity electromagnetic pulse interference generated by ultra-high-voltage transmission towers, the specific logical operation and physical evolution process is as follows: When the drone enters the core area of ​​the high-voltage transmission line at a cruising speed of 20 m / s, the high-intensity electromagnetic pulse causes the output signals of the satellite positioning unit and visual measurement unit in the data sensing module to generate high-frequency random oscillations. The spatial position feedback quantity undergoes a sudden displacement change within a 10 ms sampling period. The control processing module retrieves the pre-stored total mass parameters of the aircraft in the onboard memory in real time. and the maximum thrust-to-weight ratio parameter of the power system According to the formula Calculate the true acceleration limit threshold that an aircraft can reach when moving in three-dimensional space. ;in, The true acceleration limit threshold, This is the maximum combined thrust output by the power system. The total mass parameter of the aircraft, this threshold serves as a kinematic rigid boundary constrained by the physical inertia of the airframe, used to identify false feedback. The control processing module receives the spatial position feedback output by the data sensing module and uses second-order finite difference to calculate the observed acceleration parameter within the current sampling period. ,in, To observe the acceleration parameter, when the Exceeding two consecutive sampling periods Within a 1.2 times range of values, the control processing module determines that the feedback offset is a non-physical distortion of the sensor rather than a real physical displacement.

[0042] Under this interference condition, if the system maintains the traditional adjustment loop operation, the attitude closed-loop control logic will respond to this false large deviation, causing a step jump in the duty cycle command output by the control processing module, driving the power unit to reach its speed limit, and inducing the aircraft's attitude collapse. This solution determines... Exceeding In the transient state, the control logic directly sets the weight coefficients of the spatial position feedback and velocity feedback in the fusion adjustment algorithm to 0, thereby truncates the feedback path before the physical signal is converted into power drive. Simultaneously, the control processing module locks the error integral term in the attitude adjustment logic to the historical calculation value of the sampling period before the interference judgment, thus suppressing integral saturation. This achieves a solution to the contradiction between the need for rapid response and spurious signal interference within a single adjustment architecture. During the period when the feedback channel is disconnected, the control processing module retrieves the attitude angular rate data output by the built-in inertial measurement unit and uses it as a feedforward compensation reference input to the attitude adjustment logic, maintaining the output of the drive adjustment module. When the duty cycle command is given, the control processing module uses a preset strapdown inertial navigation solution matrix to update the attitude angular rate data using quaternion integration to lock the transient three-dimensional attitude angle of the aircraft. Using this fixed attitude constraint superimposed with the locked error integral term thrust component, the aircraft is forcibly reduced from spatial closed-loop displacement tracking to open-loop constant velocity vector translational motion within the disconnected short prediction window. This constructs an aerodynamic alternative path that maintains the inner loop angular rate for external spatial translation. Since this compensation benchmark uses high-confidence angular rate inertial data to replace the failed position feedback, the drive adjustment module drives the power unit to maintain the steady-state aerodynamic state before the disturbance occurred.

[0043] During this period, the control processing module extracts the target multimodal features acquired by the visual measurement unit and calculates the target's trajectory within the next 100ms based on the trajectory prediction logic, updating the UAV's offline flight trajectory in real time. This allows the aircraft to maintain its pointing towards the inspected target without feedback. In practice, the aforementioned image multimodal features focus on the geometric centroid coordinates of the target's edge contour and its corresponding pixel-level optical flow vector. The control processing module multiplies the optical flow vector by a pre-stored camera intrinsic focal length scaling factor, converting it into a relative angular velocity parameter in the camera coordinate system. This parameter is then substituted into a linear kinematic equation containing a first-order extrapolation operator to calculate the spatial target trajectory within the specified prediction time window along the tangent direction of the feature centroid's movement. When the UAV flies away from the strong magnetic field area, the newly calculated observation acceleration parameter... Falling back to the true acceleration limit threshold Within this timeframe, the control processing module detects that the feedback signal has regained compliance with the kinematic physical constraints, and then resumes the closed-loop feedback adjustment of the feedback channel. The weighting coefficients are restored to their normal calibration values. At this point, since the error integral term is frozen during the disturbance, the system avoids control overshoot caused by the accumulation of false deviations and re-enters the online closed-loop mode with a stable initial physical state. This process achieves a seamless control transition between online adjustment and offline tracking by taking over the inertial feedforward of the physical boundary. The UAV completes the inspection task without attitude overturning. The system's non-electric variable adjustment loop achieves the expected anti-interference control under extreme disturbance inputs, demonstrating the system architecture's adaptability to changes in the physical environment.

[0044] Example 2: In a physical simulation environment traversing a UAV performing a power line inspection mission, the test platform includes a closed flight control room with six-degree-of-freedom motion capture capability and an electromagnetic pulse generator with a maximum radiation intensity of 500V / m. The original flight state data comes from the flight recorder built into the onboard control processing module. The recorder has a data acquisition resolution of 16 bits and a position coordinate sampling rate of 100Hz. Since the system operates under dynamic inspection conditions at 20m / s, to meet the accuracy requirements for capturing transient interference and avoid discretization errors in finite difference calculations, the sampling period is... The timeframe is set to 10ms. This value ensures that at least 10 data points of effective fitting sequence are obtained within a 100ms trajectory prediction window. The experimental group adopted the scheme of this invention. When the electromagnetic pulse generator outputs electromagnetic interference of 100V / m, the spatial position feedback measured by the system exhibits waveform jitter. The acceleration parameters are observed... The true acceleration limit threshold was not reached. The system maintains online closed-loop adjustment with an attitude angle error within 1.2°, and when the electromagnetic interference intensity increases to 300V / m, the spatial position feedback output by the data sensing module changes abruptly. The control processing module then calculates... Exceeding within two consecutive sampling periods The control logic then sets the weight coefficients of the spatial position feedback and velocity feedback in the fusion adjustment algorithm to 0, and the attitude angle rate feedforward compensation logic takes over the driving adjustment module. Its attitude angle fluctuation peak is 3.8°. In contrast, the control group that did not adopt the error integral term locking feature, under the same working conditions, caused the duty cycle command of the power unit to reach 100% saturation within 50ms due to the continuous accumulation of false deviations in the integral term, resulting in a 15.5° overturning deviation of the aircraft and an oscillation duration of more than 2.5s. This data confirms that the feedback channel weight adjustment and the error integral locking mechanism have a synergistic effect in suppressing feedback distortion.

[0045] To determine the numerical range boundaries, an out-of-range control group with a threshold coefficient of 3.0 was set. Under electromagnetic interference conditions of 500V / m, due to the excessively wide coefficient setting, the system failed to identify non-physical distortions, leading to abnormal position feedback intrusion into the control loop. This caused a 45% step increase in the average thrust output of the drive adjustment module, inducing a stall. Another out-of-range control group with a threshold coefficient of 0.5 exhibited misjudgment during normal high-overload maneuvers, resulting in the disconnection of the position feedback channel and an increase in the trajectory tracking error of attitude adjustment to 0.82m. Meanwhile, the attitude angle deviation of the experimental group at a threshold coefficient of 1.2 increased linearly with the interference intensity from 100V / m to 500V / m. When the interference was removed and... Falling back to Once within range, the position feedback channel recovers, and the adjustment loop converges within 80ms, confirming that the 1.2 times threshold configuration is a working window that balances anti-interference sensitivity and maneuver reliability; this is achieved by anchoring the observed acceleration parameters to the physical kinematic boundaries of the aircraft. Above, among which, The true acceleration limit threshold, This is the maximum combined thrust output by the power system. The total mass parameter of the aircraft can convert unpredictable electromagnetic interference into a deterministic acceleration envelope verification process. When the system faces sensor feedback distortion, it uses the error integral register value constant processing method to block the accumulation of false deviations, so that the duty cycle command is kept within the physical safety envelope range determined by the maximum thrust-to-weight ratio parameter. This ensures that the aircraft maintains its pointing towards the target in the absence of feedback and achieves smooth recovery of the adjustment loop. The non-electric variable adjustment loop of the system has engineering stability under multiple interference conditions.

[0046] Example 3: This example combines Figures 1 to 2 This describes a type of unmanned aerial vehicle (UAV) system with offline anti-interference and dynamic intelligent tracking capabilities, such as... Figure 1 As shown, the total mass parameters and maximum thrust-to-weight ratio parameters of the aircraft are obtained to initialize the inherent physical constraint parameters of the airframe, and the true acceleration limit threshold is calculated to establish the physical boundary of the three-dimensional spatial motion. Simultaneously, the motion velocity data provided by the output velocity feedback and the spatial position data provided by the output spatial position feedback are used to calculate the observed acceleration parameters and extract second-order or first-order finite difference values. These values ​​are then entered into the front-end validity verification stage to determine whether the observed acceleration parameters are greater than the true acceleration limit threshold and whether they meet the judgment condition for two consecutive sampling periods. If the judgment result meets the preset conditions, it is determined to be non-physical. In case of distortion, the system disconnects the feedback channel and locks the error integral term value, setting the weight coefficient to 0 to suppress integral saturation. At this time, the attitude angular rate data output by the built-in inertial measurement unit is retrieved and used as the feedforward compensation reference for feedforward compensation takeover. Under the logical guidance of maintaining a smooth duty cycle command, the system receives the duty cycle command. If the front-end legality check result is negative, otherwise it is determined to fall back to the threshold. The system resumes closed-loop feedback regulation and makes the feedback state variable fall back to the physically reasonable range. Finally, the speed control unit receives the duty cycle command to maintain stable aerodynamic attitude and power unit speed.

[0047] like Figure 2As shown, the aerial drone system comprises a perception hardware platform, an airborne control core platform, and a power drive terminal in terms of hardware and logic architecture. The perception hardware platform integrates a satellite positioning unit, a visual measurement unit, and a built-in inertial measurement unit. An interference suppression and isolation module suppresses the built-in inertial measurement unit through physical signals. The perception hardware platform interacts with the airborne control core platform through a high-frequency signal in the data feedback channel. The airborne control core platform is equipped with a multi-source data fusion algorithm, attitude adjustment logic, an error integral register, a feedforward compensation mapping table, and a trajectory prediction logic. These are used to perform logical calculations on the input perception data and generate physical adjustment link duty cycle commands. These commands are output to the power drive terminal to control its internal speed control unit, thereby driving the power unit to operate.

[0048] Example 4: To determine the multiplier factor for non-physical distortion of the feedback path, flight maneuver calibration is performed in a non-interference environment beforehand, and the control processing module calls the maximum thrust-to-weight ratio parameter of the power system. Drive the drone to perform extreme overload maneuvers and collect and observe acceleration parameters in real time. Compared with the true acceleration limit threshold The deviation value is calculated, and the 99% confidence interval boundary of the deviation value is added to this boundary. The sensor noise redundancy was calculated, and a decision threshold of 1.2 times was ultimately determined. For a 100ms trajectory prediction window, alignment was performed based on the 100Hz sampling frequency of the visual measurement unit to ensure that the prediction period includes 10 visual sampling frames. During the transient process of the UAV passing through the canopy, the ambient light intensity suddenly drops from 50,000 lux to below 100 lux, causing the target feature vector output by the visual measurement unit to... Semantic loss occurred, and the control processing module detected it. If the value exceeds the 1.2 times threshold value determined by calibration for two consecutive sampling cycles, it is determined that the feedback path has a non-physical distortion. The control processing module then sends an interrupt instruction to the attitude control register of the airborne control chip, triggering the hardware logic gate circuit to cut off the input bus path of the integrator accumulator, so that the value in the error integration register remains in the solidified value state of the previous instruction cycle before the interference judgment, preventing the ill-conditioned deviation from entering the integration operation unit.

[0049] During the task takeover period when the feedback channel weight is set to 0, the control processing module retrieves the historical coordinate sequence within the 5 sampling periods prior to the visual loss. With attitude angular rate data Using the first-order Taylor expansion operator pair Perform differentiation to obtain the target relative velocity vector, and then correlate it with the sampling period. After accumulating the points, according to A coordinate system rotation transformation is performed to generate a target pointing vector representing the future trajectory. This pointing vector is mapped to the desired attitude angle command of the drive adjustment module to maintain the pointing stability of the UAV in the absence of external position feedback. When the UAV flies away from the light and shadow interference area, allowing the visual measurement unit to recapture effective features, a new calculation is performed. Falling back to Within the constraints, the control processing module withdraws the interrupt command and the input bus of the integral accumulator is reconnected by the hardware logic gate circuit, and the weighting coefficient is restored from 0 to the preset calibration value. Since the integral register is in a physically locked state during the interference, the duty cycle command output by the drive adjustment module is restored to steady-state aerodynamic balance within 15ms, realizing the physical self-consistency and lossless recovery of the non-electric variable adjustment loop under the condition of perception failure. The UAV maintains the established inspection track in the complex forest environment.

[0050] Example 5: In the power performance calibration condition, the UAV power unit is vertically fixed to the force-receiving end of a high-precision force sensor with a range of 0N to 200N and a sampling rate of not less than 1000Hz. The control processing module sends a step-wise pulse width modulation signal through the drive adjustment module. When the power unit reaches the thrust saturation state under the maximum duty cycle command, the steady-state maximum combined thrust output by the sensor is recorded. ,in, The maximum combined thrust output of the propulsion system is measured, while the total mass parameters of the aircraft are read from the electronic balance. The control processing module calculates the maximum combined thrust. With the total mass parameters of the aircraft The quotient is calculated and stored in the onboard memory to determine the true acceleration limit threshold characterizing the kinematic boundary of the body. .

[0051] When the drone is operating at varying altitudes, the data sensing module collects data from the barometer and onboard temperature sensor in real time. The control processing module calculates the density correction factor of the current ambient density relative to standard atmospheric pressure based on the atmospheric density model, and then applies this density correction factor to the true acceleration limit threshold. The system performs corrections while simultaneously monitoring the residual distribution of the motor's attitude angular rate under no-load rotation. It calculates the noise reference value of the inertial measurement unit in the current environment and injects this noise reference value as an additive term into the prediction step size of the attitude adjustment logic. This allows the non-electric variable adjustment loop to complete parameter recalibration based on changes in the environmental physical state. The atmospheric density model essentially follows the ideal gas state law. It calculates the current high-altitude absolute density by dividing the real-time acquired air pressure value by the product of the onboard temperature and a specific gas constant. The control processing module divides this absolute density by the standard atmospheric density at sea level to obtain a dimensionless correction factor, which is then used as a linear multiplier to directly scale the limit threshold proportionally based on the predefined limit thrust-to-weight ratio reference to compensate for the overall thrust reduction caused by thin air.

[0052] In the underlying mapping scenario of the UAV power system, the control processing module retrieves the rotor lift coefficient pre-stored in the onboard memory. and the upper limit of motor speed The duty cycle signal output by the attitude control logic Through a preset physical mapping function This is converted into thrust parameters; while the error integral term is locked, the drive adjustment module receives the fixed integral component and combines it with real-time attitude angular rate data, using a limiting operator to restrict the output duty cycle command to the maximum thrust-to-weight ratio parameter of the power system. Within the defined linear response range of 0.15 to 0.95, the duty cycle command output by the control processing module always corresponds to the physical effective thrust range of the power unit, avoiding the risk of rotor anti-drag stall or ESC overload caused by control signal step changes. In the task initialization scenario of performing multi-target intelligent tracking, the control processing module implements a normalization preprocessing procedure at the input of the multi-source data fusion algorithm, converting the global coordinate data output by the satellite positioning unit... The target feature vector output by the visual measurement unit Mapped to a unified body coordinate system space, the initial fusion weights of the satellite signals are set as follows: And the initial fusion weights of the visual features are This multi-source data fusion algorithm deploys a Kalman filter mathematical architecture, and the control processing module applies the aforementioned initial fusion weights. and The inverses of the diagonal elements of the measurement noise covariance matrix of the satellite positioning submodule and the vision submodule in the measurement update equation are respectively configured. By solving for the Kalman gain and multiplying it with the multi-source observation residual, the pose state prediction vector output by the kinematic prior model is corrected. When the system detects the observed acceleration parameter... Exceeding the true acceleration limit threshold When the weight is reset to zero, the control processing module will... or Set the value to 0 and send a state freeze flag to the fusion adjustment algorithm. This flag triggers the algorithm to maintain the covariance matrix at the value before the interference judgment until the feedback signal passes the kinematic boundary verification again and the weight is restored stepwise according to the preset time constant.

[0053] Example 6: In the static zero-point calibration scenario before aircraft takeoff, the control processing module acquires the raw angular velocity sequence of the inertial measurement unit in a powerless output state. Calculate the standard deviation of the sequence within a 2000ms time window. Then, taking three times the standard deviation, this calculation result is defined as the zero-bias drift suppression threshold in the dynamic adjustment logic, thereby determining the physical sensitivity boundary of the attitude angular rate feedforward compensation logic; simultaneously, the target feature vector output by the vision measurement unit is considered. The control and processing module implements the feature saliency distribution calibration procedure under controlled lighting conditions, calculates the contrast of the input image using a histogram equalization operator, extracts the gradient magnitude distribution of the 16×16 pixel block around the feature point, and sets the confidence weight for determining the existence of the target. The initial value is 0.85 to ensure that the multi-source data fusion algorithm has a defined quantization calibration benchmark during operation; among which, This is the original angular velocity sequence. Standard deviation For the target feature vector, As a confidence weight, the initial benchmark of 0.85 is determined based on the static quantization deduction during the system's factory testing phase. The control processing module pre-inputs the template grayscale array of the inspection target under the basic illumination environment. Within the algorithm, this interference-free template is regarded as the ideal judgment benchmark. The similarity parameters between the distribution of the currently collected feature pixel blocks and the standard template are analyzed by the normalized cross-correlation operator. By filtering out abnormal test points that cause semantic breaks due to structural occlusion, the similarity critical empirical value for maintaining the visual loop closure limit in engineering statistics is extracted and set to 0.85.

[0054] When the system is in an offline takeover state with the feedback channel weights set to zero, the control processing module runs a trajectory prediction procedure based on a constant acceleration model, retrieving the spatial position feedback from the data sensing module during the last valid sampling period before the interference occurred. With speed feedback quantity The expected displacement within a 100ms prediction step is calculated using a second-order Taylor series expansion model, and the expected value of the target position is obtained. The calculation method is to use With sampling period The product of these terms is used as a first-order component and is combined with the observed acceleration parameter. one-half and The summation of squared product terms, in the logic for determining the recovery of online closed-loop regulation, the control processing module sets a debouncing window containing 5 consecutive sampling periods. Maintain the true acceleration limit threshold within the debouncing window. When the value is within 0.8 times the original value, the system triggers the hardware logic gate circuit to reconnect the input bus of the integral accumulator, and the weighting coefficient is restored from 0 to the preset calibration value. This avoids logic switching oscillations caused by residual interference. After the physical parameters are recalibrated, the attitude control loop of the UAV system enters a stable operating state.

[0055] 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.

[0056] 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 drone system with offline anti-interference and dynamic intelligent tracking capabilities, characterized in that, It includes a data sensing module, a drive adjustment module, and a control processing module: The control processing module is connected to the data sensing module and the drive adjustment module via signals, respectively. The control processing module is used to obtain the total mass parameters of the aircraft and the maximum thrust-to-weight ratio parameters of the power system, and to calculate the actual acceleration limit threshold that the aircraft can reach when moving in three-dimensional space. The control processing module is used to receive the spatial position feedback and velocity feedback output by the data sensing module, calculate the second-order finite difference value of the spatial position feedback within the current sampling period, or calculate the first-order finite difference value of the velocity feedback, and determine the observation acceleration parameter that characterizes the actual observation state. The control processing module is used to implement the front-end legality verification of the adjustment loop. When the observed acceleration parameter is greater than the true acceleration limit threshold for two consecutive sampling periods, it determines that the feedback path has a non-physical distortion. The weight coefficients of the spatial position feedback and velocity feedback in the fusion adjustment algorithm are set to 0 to disconnect the feedback channel. At the same time, the error integral term in the attitude adjustment logic is locked to the historical calculation value of the sampling period before the interference judgment to suppress the integral saturation phenomenon caused by feedback distortion. The control processing module is used to retrieve the attitude angular rate data output by the built-in inertial measurement unit during the period when the feedback channel is disconnected, and use it as the feedforward compensation reference input to the attitude adjustment logic to maintain the smooth duty cycle command output of the drive adjustment module until the newly calculated observed acceleration parameter falls back to within the true acceleration limit threshold, and then the closed-loop feedback adjustment of the feedback channel is restored.

2. The unmanned aerial vehicle system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The control processing module is also used to keep the corresponding error integral register value constant during the period of locking the error integral term value, so as to block the accumulation of false deviations caused by feedback signal distortion, and keep the duty cycle command output by the control processing module within the physical safety envelope determined by the maximum thrust-weight ratio parameter, so as to ensure that the aircraft maintains the original aerodynamic attitude during the offline mission logic takeover transient, and avoids the rotor overshoot surge caused by control signal overload.

3. The unmanned aerial vehicle system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The data perception module includes a satellite positioning unit and a visual measurement unit; the control processing module is used to determine the dynamic intelligent tracking path of the UAV relative to the target through a multi-source data fusion algorithm based on the global coordinate data provided by the satellite positioning unit and the target feature vector provided by the visual measurement unit; the control processing module verifies the input parameters entering the multi-source data fusion algorithm through the real acceleration limit threshold in each adjustment cycle.

4. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The drive adjustment module includes a speed control unit and a power unit; the duty cycle command output by the control processing module is used to drive the speed control unit; when the feedback channel is disconnected, the control processing module controls the speed control unit by adjusting the feedforward compensation reference to maintain the stable speed of the power unit.

5. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The control processing module is also used to extract multimodal features of the image acquired by the vision measurement unit and calculate the target's trajectory within the next 100ms based on the trajectory prediction logic. When the feedback channel is disconnected, the control processing module updates the UAV's offline flight trajectory in real time based on the trajectory, so that the aircraft can maintain its pointing towards the target in the absence of feedback.

6. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The control processing module is also used to monitor the signal quality parameters of the data sensing module. When the signal quality parameters are below 5dB and the duration exceeds 50ms, the control processing module switches to offline cruise mode and generates control signals to drive the adjustment module by reading the mission waypoint data in the built-in storage module and combining it with the attitude angular rate data.

7. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The control processing module stores a feedforward compensation mapping table, which records the basic duty cycle data corresponding to different flight speed ranges. During the period when the feedback channel is disconnected, the control processing module retrieves the corresponding basic duty cycle data according to the current flight speed and adds the deviation correction amount generated by the attitude angular rate data.

8. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The system also includes a communication link monitoring module, which is used to detect the continuity of external command signals; When the feedback channel is disconnected and the external command signal is interrupted, the control processing module activates the offline autonomous logic and uses the steady-state motion parameters before the feedback channel was disconnected as initial values ​​to calculate the flight trajectory.

9. A drone system with offline anti-interference and dynamic intelligent tracking as described in claim 1, characterized in that, The system also includes an interference suppression and isolation module. The data sensing module is deployed at the center of the aircraft through the interference suppression and isolation module to attenuate the interference of mechanical vibration generated by the drive adjustment module on the acquisition of spatial position feedback.

Images