An unmanned aerial vehicle active slow descent control method and system based on dynamic thrust deficit and a medium

Abstract

The application provides a kind of unmanned aerial vehicle active slow descent control method, system and medium based on dynamics thrust deficit;In view of the technical pain point that unmanned aerial vehicle is caused by sensor pollution in complex working conditions Protection device false triggering, and slow descent equipment is easy to be entangled and cut off by high-speed rotor when being thrown, the application first monitors the time consistency of multi-source sensing data, actively isolates the disturbed data and cuts into the physical inertia takeover to prevent strong electromagnetic interference;At the same time, a supply-demand game model between the demand of kinetic destruction thrust and the physical limit of underlying motor is constructed, and the irreversible falling state is accurately judged in combination with safety height to prevent early explosion caused by attitude disturbance;Finally, after the right to fall, the underlying power output is cut off and the anti-winding deceleration timing interlocking mechanism is introduced, and the slow descent device is triggered after the rotor kinetic energy is safely attenuated, realizing the closed loop of falling right and safe recovery, and improving the survival rate of flight platform.

Description

Technical Field

[0001] This invention relates to the field of active safety protection for unmanned aerial vehicle (UAV) flight control, and in particular to an active descent control method, system, and medium for UAVs based on dynamic thrust deficit. Background Technology

[0002] With the widespread application of drones in complex operational scenarios such as power line inspection and high-rise building monitoring, their flight safety in environments with strong electromagnetic interference and complex wind fields has become a core focus of industry attention. When a drone encounters sudden attitude instability or mechanical power failure, accurately assessing the fall status and safely triggering the descent protection device is crucial to preventing damage to high-value flight platforms and secondary safety accidents on the ground.

[0003] In the prior art, multi-sensor data fusion is usually used to maintain the attitude stability of UAVs. For example, Chinese patent publication CN104914874A discloses "A UAV Attitude Control System and Method Based on Adaptive Complementary Fusion", which proposes to acquire data from accelerometers, gyroscopes and magnetometers, adjust the confidence weight of each sensor through an adaptive complementary filter, and then calculate the attitude angle of the UAV, and adjust the motor output accordingly to correct attitude deviation.

[0004] However, when faced with complex and harsh real-world working environments, existing technologies, represented by this particular technology, reveal the following technical shortcomings: First, multi-source fusion in strong magnetic field environments is highly susceptible to data contamination, lacking discrepancies and rigid isolation mechanisms at the physical time-series level. Existing technologies rely excessively on adaptive weight adjustment at the software algorithm level. When a drone approaches strong magnetic field areas such as high-voltage power lines, the environmental magnetic field data output by the magnetometer will be distorted. Existing fusion mechanisms cannot identify the heterogeneous data discrepancies between magnetic field abrupt changes and angular velocity / acceleration abrupt changes caused by actual body fall from the physical underlying time series and rigid body kinematics. This leads to the system easily mistaking magnetic field distortion for body roll, thus inputting a contaminated attitude matrix into the control calculation and causing drone control divergence. Secondly, pure attitude correction logic is detached from the underlying physical limits and cannot perform dynamic assessment of irreversible crash conditions. The control logic of existing technology only stays at the surface level of attitude deviation correction and lacks calculation of the underlying mechanical dynamic limits of the UAV. When the UAV encounters extreme downdrafts or partial failure of the power system, simple attitude angle error feedback cannot assess the current physical recovery capability of the aircraft. The system cannot calculate the real destructive thrust required to overcome the current descent kinetic energy, nor can it compare the thrust deficit between this requirement and the maximum full-load physical thrust limit of the motor. This leads to the existing system still blindly outputting full throttle commands to try to recover attitude in real irreversible crash conditions, which not only fails to assess the fall state but also completely misses the best safe altitude and time to trigger the protection device. Furthermore, the decoupling of the terminal protection action from the underlying power system poses an extremely high risk of secondary disasters caused by mechanical entanglement. Existing flight control architectures, represented by current technologies, lack a rigid physical interlock design for the execution path of the underlying hardware under extreme conditions. Even if the superstructure is equipped with parachutes or other descent control devices, the system cannot forcibly cut off the high-voltage power supply to the electronic governor in a runaway, fully loaded state on both the physical and logical links when the parachute ejection command is triggered. Due to the extremely fast transmission of electrical signals and the physical delay in the deceleration of mechanical rotors, the lack of a forced waiting time for the mechanical kinetic energy decay dead zone can easily lead to the ejected parachute lines or airbags being instantly cut off or entangled by the still-spinning propellers, causing the descent protection to completely fail and triggering a serious secondary crash.

[0005] In response to the aforementioned technical deficiencies such as the susceptibility of sensing data to contamination, the lack of irreversible physical drop confirmation, and the absence of timing-based anti-entanglement interlocks in hardware descent execution, there is an urgent need in this field for an active descent control method and system for unmanned aerial vehicles that is directly based on physical logic for game theory and possesses timing and strong electrical physical interlocking mechanisms. Summary of the Invention

[0006] The main objective of this invention is to provide an active descent control method, system, and medium for unmanned aerial vehicles (UAVs) based on dynamic thrust deficit. Existing technologies are prone to sensor data contamination and false alarms in complex operating environments such as strong magnetic fields, lack a fundamental physical and dynamic determination mechanism for the irreversibility of the aircraft's fall, and suffer from technical problems such as the disconnect between the descent device trigger command and the underlying rotor power, leading to parachute lines being cut or entangled by high-speed blades and failing.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an active descent control method for unmanned aerial vehicles based on dynamic thrust deficit, comprising the following steps: S1. Obtain raw angular velocity data Raw magnetic field data Compared with raw acceleration data Calculate and generate background magnetic field gradient reference values With the initial attitude transformation matrix In the raw acceleration data Exceeding the limit, and based on raw angular velocity data With the original magnetic field data Calculated heterogeneous data deviation coefficient When the limit is exceeded, output a one-way wake-up flag. ; S2, Obtain the one-way wake-up flag. Then, based on the real-time magnetic field gradient Reference value of background magnetic field gradient The difference will be used to integrate the magnetometer weighting coefficient. Set to zero; during the safe redundancy inertial period Internally, based on the initial attitude transformation matrix Angular velocity raw data Compared with raw acceleration data Perform continuous integration to generate a real-time attitude transformation matrix. ; S3, Based on real-time attitude transformation matrix Perform dynamics calculations to generate the existing failure thrust requirements. Based on the motor's maximum instantaneous full-load thrust limit parameters Compared with existing destructive thrust requirements The thrust difference and the real-time absolute altitude above the ground Perform a graded judgment and output an irreversible fall confirmation signal when the irreversible fall condition is met. ; S4. Receive irreversible drop confirmation signal Then, a cutoff signal was sent. And disconnect the main power supply circuit to generate a power disconnection confirmation flag. ; Confirmation of power cut-off flag Start-up mechanical deceleration dead time A countdown timer will run out, after which an action trigger command will be output. Drive the descent actuator and store the fault snapshot dataset. .

[0008] In the preferred embodiment, step S1 specifically includes: Extract raw acceleration data Instantaneous value of vertical axis acceleration in ; Instantaneous value of vertical axis acceleration Maximum Redundancy Lift Compensation Limit Comparative evaluation was conducted; the instantaneous value of the vertical axis acceleration was determined. Breaking the maximum redundancy lift compensation limit When this happens, an internal timer is triggered to open the initial monitoring time window. and in the initial monitoring time window Internal activation heterogeneous sensor timing coupling deviation verification mechanism.

[0009] In the preferred embodiment, step S1 further includes: Extract raw angular velocity data separately Dynamic mutation rate and original magnetic field data The rate of change of heading; combined with the static proportional mapping coefficient Dimensional alignment was performed on the dynamic mutation rate and the rate of yaw change to assess the physical coupling consistency between the dynamic mutation rate and the rate of yaw change in the kinematics of the rigid body in space, and the heterogeneous data bias coefficient was determined. ; the deviation coefficient of heterogeneous data Physical coupling limit threshold A comparison was performed to determine the deviation coefficient of the heterogeneous data. Exceeding the physical coupling limit threshold At that time, a one-way wake-up flag is generated. Forced entry into the irreversible drop verification state machine.

[0010] In the preferred embodiment, step S2 specifically includes: Comparison of real-time magnetic field gradients Reference value of background magnetic field gradient The gradient drift is acquired, and when the gradient drift exceeds a preset step threshold, a physical-level confidence assignment is triggered to lock the magnetometer fusion weight coefficients in the multi-source sensing fusion model. Zero; during the safe redundancy inertial period when timing is activated. Inside, shield the magnetic field heading reference data and use the initial attitude transformation matrix. Using this as the starting point for calculation, combined with the original acceleration data... Extracted raw data of gravity vector and angular velocity Perform continuous integration of the rigid body space differential equations to generate a real-time attitude transformation matrix. .

[0011] In the preferred embodiment, step S3 specifically includes: Based on real-time attitude transformation matrix Angular velocity raw data and raw acceleration data Extracting the downward acceleration component Yaw angular momentum The extracted fall acceleration components were obtained based on the overall mass of the drone. Mapped to the initial thrust required to overcome the downward fall, and the yaw momentum of the airframe combined with the arm moment parameters. This is converted into a second thrust requirement to suppress the aircraft's roll; a vector synthesis process is performed on the first and second thrust requirements to generate the existing destructive thrust requirement. .

[0012] In the preferred embodiment, step S3 further includes: Compare the maximum instantaneous full-load thrust limit parameters of the motor Compared with existing destructive thrust requirements Generate thrust compensation difference that characterizes the dynamic supply and demand boundary. ; in thrust compensation difference It is in a negative state and continues to exceed the transient gust filter window. When the attitude is determined to be in an irrecoverable state, the ranging sensing hardware is activated in parallel to read the real-time absolute altitude above the ground. Real-time absolute altitude above the ground Minimum safety distance from umbrella opening Comparison, to confirm the real-time absolute altitude above the ground Falling below the minimum safety distance for parachute opening When this occurs, an irreversible drop confirmation signal is output in a high-level state. .

[0013] In the preferred embodiment, step S4 specifically includes: Capture irreversible fall confirmation signal Subsequently, an anti-entanglement interlock mechanism is triggered synchronously on both the logical and physical links; the communication control bus enable pin is pulled low on the logical link, and a zero-width truncation signal is sent. The motor drive duty cycle is reset to zero; a circuit breaker command is sent to the main power supply circuit in parallel on the physical link to cut off the high-voltage supply to the electronic speed controller; after confirming that the logical link and physical link are blocked and the underlying power output is cut off, a power cut-off confirmation flag is generated. .

[0014] In the preferred embodiment, step S4 further includes: Based on the power cut-off confirmation flag The pre-calibrated mechanical deceleration dead time An independent hardware timer is loaded to execute the countdown, and the control bus of the deceleration actuator is locked to wait for the rotor's mechanical kinetic energy to cause air resistance decay; during the mechanical deceleration dead time... When the countdown expires, confirm the dead zone exhaustion flag. When in an active state, the closed-end drive unit circuit outputs an action trigger command. Activate the igniter or solenoid valve; acquire the three-dimensional spatial coordinates of the aircraft's position and the state machine transition characteristic parameters at the moment of the crash in parallel, and package them to construct a fault snapshot dataset. It is then written to the external flash memory black box.

[0015] In a preferred embodiment, to achieve the above objectives, the present invention also provides an active descent control system for unmanned aerial vehicles based on dynamic thrust deficit, comprising: The multi-source timing wake-up unit is used to acquire raw sensing data of angular velocity, magnetic field, and acceleration, and to calculate the background magnetic field gradient reference value. With the initial attitude transformation matrix ; and the deviation coefficient between acceleration components and heterogeneous data When both limits are exceeded, output a one-way wake-up flag. ; Physical stripping takeover engine for responding to one-way wake-up flags The magnetometer is integrated with weighting coefficients based on the magnetic field gradient difference. Set to zero; and during the safe redundancy inertial period Perform continuous integration operations to generate a real-time attitude transformation matrix. ; The thrust deficit determination module is used to determine the thrust deficit based on the real-time attitude transformation matrix. Solving for existing destructive thrust requirements Combined with the motor's full-load thrust limit parameters Real-time absolute altitude above the ground Perform a drop test and output an irreversible drop confirmation signal. ; Timing interlock execution terminal, used to respond to irreversible drop confirmation signal. Send cutoff signal And disconnect the power supply circuit to generate a power disconnection confirmation flag. ; and during the mechanical deceleration dead zone time After the countdown expires, output the action trigger command. Drive the slow descent actuator and store the fault snapshot dataset. .

[0016] To achieve the above objectives, the present invention further provides a computer-readable storage medium that stores computer-executable instructions, which, when read and executed by a processor, implement the active descent control method for unmanned aerial vehicles based on dynamic thrust deficit as described above.

[0017] This invention provides an active descent control method, system, and medium for unmanned aerial vehicles (UAVs) based on dynamic thrust deficit. By constructing a dynamic physical perception and state verification mechanism, it improves the survivability of UAVs under complex working conditions. Specifically, this invention reconstructs the single gravity acceleration limit judgment into a dynamic verification process that combines the rotor's maximum redundant lift compensation limit, and introduces a timing deviation verification mechanism for multi-source sensors. When the UAV encounters a strong magnetic field from a high-voltage coil during power line inspection, the system can keenly detect the step change in the magnetic field gradient, actively reset the fusion confidence weight of the magnetometer to zero, and switch to a short-term physical takeover mode within the safe redundancy inertial period. This mechanism, from a kinematic logic perspective, blocks the malicious interference of distorted data from strong electromagnetic pollution on the flight control attitude calculation matrix, thereby reducing the risk of false triggering of protection devices in complex electromagnetic fields from the source. Based on this, the present invention also visualizes the attitude optimization black box at the algorithm level as a dynamic thrust deficit calculation game with logical significance. In the face of extreme descent conditions, the system can calculate in real time the thrust required to overcome the current destructive kinetic energy of the aircraft and compare it with the instantaneous full-load physical thrust limit that the underlying electronic speed controller can call. Only when the physical recovery limit of the system is completely unable to cover the current destructive kinetic energy, and the thrust deficit state has undergone the verification of the gust transient filtering time window and the double hard constraint of the safety height of parachute opening after takeoff, does the system logically finally confirm that the deviation is an irreversible real crash state. This gives the UAV a very high system fault tolerance space, filters the high maneuvering attitude disturbance caused by actively avoiding obstacles or responding to sudden wind shear, and alleviates the premature detonation and malfunction of the descent actuator. Furthermore, addressing the issue that parachutes or airbags are easily severed by the still-high-speed propeller during launch, this invention constructs an extremely robust anti-entanglement physical interlock timing defense at the execution end. Before finally issuing the descent trigger command, the underlying control interlock is forcibly activated, the duty cycle signal is simultaneously forcibly cleared to zero on the logic link, and the main circuit power supply is cut off on the physical link. At the same time, a deceleration dead zone countdown mechanism based on the rotor's mechanical rotational inertia is introduced. Relying on an independent hardware timer, it is strictly ensured that the rotor's mechanical kinetic energy has undergone substantial physical attenuation under the action of air resistance. Only after completely crossing the mechanical deceleration dead zone is the system released the execution authority of the end drive circuit, making up for the time difference between electrical signal transmission and mechanical component deceleration, eliminating the risk of secondary disasters such as the descent equipment being damaged by its own rotor, ensuring the physical safety of high-value inspection assets and ground facilities, and leaving highly reliable traceability data for subsequent fault tracing. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of an active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to the present invention; Figure 2 This is a schematic diagram of an active descent control system for unmanned aerial vehicles based on dynamic thrust deficit, according to the present invention. Detailed Implementation

[0019] Example 1 like Figures 1 to 2 As shown, an active descent control method for unmanned aerial vehicles (UAVs) based on dynamic thrust deficit includes the following steps: S1. Obtain raw angular velocity data Magnetic field raw data Compared with raw acceleration data Calculate and generate background magnetic field gradient reference values With the initial attitude transformation matrix In the raw acceleration data Exceeding the limit, and based on raw angular velocity data With the original magnetic field data Calculated heterogeneous data deviation coefficient When the limit is exceeded, output a one-way wake-up flag. ; S2, Obtain the one-way wake-up flag. Then, based on the real-time magnetic field gradient Reference value of background magnetic field gradient The difference will be used to integrate the magnetometer weighting coefficient. Set to zero; during the safe redundancy inertial period Internally, based on the initial attitude transformation matrix Angular velocity raw data Compared with raw acceleration data Perform continuous integration to generate a real-time attitude transformation matrix. ; S3, Based on real-time attitude transformation matrix Perform dynamics calculations to generate the existing failure thrust requirements. Based on the motor's maximum instantaneous full-load thrust limit parameters Compared with existing destructive thrust requirements The thrust difference and the real-time absolute altitude above the ground Perform a graded judgment and output an irreversible fall confirmation signal when the irreversible fall condition is met. ; S4. Receive irreversible drop confirmation signal Then, a cutoff signal was sent. And disconnect the main power supply circuit to generate a power disconnection confirmation flag. ; Confirmation of power cut-off flag Start-up mechanical deceleration dead time A countdown timer will run out, after which an action trigger command will be output. Drive the descent actuator and store the fault snapshot dataset. .

[0020] In the preferred embodiment, step S1 includes the following steps: S11. Initialize the characteristics of the reference background field. After the drone is powered on, its built-in microelectromechanical system (MEMS) heterogeneous integrated hardware structure enters standby cruise mode; the system's underlying microprocessor calls the multi-physics data sensing module via the SPI bus to continuously acquire raw environmental background data of the inspection area where the drone is located; the heterogeneous data processing module receives raw real-time angular velocity data from the gyroscope. Real-time ambient magnetic field raw data from the magnetometer During the drone's stable hovering or undisturbed normal cruise phase, the system uses a fixed-length time sliding window algorithm to process the real-time ambient magnetic field data from the magnetometer. Spatial differentiation and mean filtering are performed to extract and calculate the baseline value of the output background magnetic field gradient. Synchronous processing of raw angular velocity data from the gyroscope in real time Statistical analysis of static drift was performed to calculate the variance of the conventional background noise of the generated gyroscope angular velocity. Simultaneously, the system combines real-time raw data from the triaxial accelerometer. The gravity vector distribution in the model is calculated using Euler angles to initialize the initial attitude transformation matrix of the UAV in the absolute space coordinate system. As the background physical constraint variables under the normal operating trajectory of the drone, they are loaded into the storage module of the high-speed flash memory; Dynamic monitoring and extreme value breakthrough determination of instantaneous drop components on S12 and Z axes During the power line inspection flight, the heterogeneous data processing module receives real-time raw data from the triaxial accelerometer. The instantaneous value of the Z-axis acceleration perpendicular to the ground plane and pointing downwards is extracted from it using the direction cosine matrix. The system internally has a pre-defined maximum redundancy lift compensation limit. This physical quantity characterizes the drop-resistance acceleration threshold that the flight control board can provide when, without triggering overall structural instability, it commands the rotor motors to output full-speed, full-load upward thrust via the electronic speed controller (ESC). The system's underlying state machine continuously adjusts the instantaneous Z-axis acceleration value at millisecond intervals. Maximum Redundancy Lift Compensation Limit Perform a comparison operation; if This indicates that the current downward vertical fall kinetic energy of the drone is within the physical control range of the rotor motors, belonging to a normal obstacle avoidance maneuver or slight gust disturbance that can be actively pulled back by the flight control system. The system maintains a normal patrol state machine and does not generate an alarm; if it detects... The system indicates that the current downward trend of the drone has completely exceeded the physical limits of the underlying rotor's compensation capabilities, and determines that there is a high risk of fall, immediately generating a fall risk flag. Simultaneously, it triggers the system's internal hardware timer to open and record an initial monitoring time window. The range is set between 50ms and 150ms to filter out single singularity anomalous data caused by instantaneous high-frequency vibration of the drone gimbal or rotor. S13. Verification of heterogeneous sensor timing coupling deviation within the initial monitoring time window The system microprocessor first checks the interrupt condition and detects the incoming drop risk flag. Subsequently, the system was limited to the initial monitoring time window. The system performs a time-series logic check to force the activation of the heterogeneous sensor; within this time window, the system calls the regular background noise variance check. As an adaptive cutoff parameter for the low-pass filter, it is used for the raw real-time angular velocity data of the gyroscope. High-frequency noise reduction is performed, followed by first-order backward differential operation to extract the comprehensive angular velocity mutation rate. Simultaneously, the system processes the raw data of the real-time ambient magnetic field from the magnetometer. The attitude deflection angle is calculated and differentiated to extract the magnetic heading rate of change. Based on the physical properties of rigid body kinematics, in the absence of strong external field interference, the physical roll of the UAV body will inevitably be accompanied by a magnetic field-induced heading deflection of the same regularity. To verify this coupling characteristic, the system calls static proportional mapping coefficients that match different physical dimensions. Construct a time series deviation test formula; Specifically, heterogeneous data deviation coefficient The dynamic calculation equation is given by the following formula: (1); in, for The deviation coefficient of the heterogeneous data generated at each time step; The combined angular velocity change rate measured by the gyroscope and after noise reduction; The rate of change of magnetic heading is sensed and derived by the magnetometer; The static scaling coefficient is used to match the difference between the dimensions of angular velocity and magnetic field deflection. The system will calculate the heterogeneous data deviation coefficient. The signal is fed into the comparator in real time and compared with a preset physical coupling limit threshold. Perform a comparison; if found If this occurs, it is determined that the drone not only experienced a physical fall, but also that the actual rolling motion of the drone is physically contradictory to the heading data sensed by the magnetometer. This phenomenon indicates that the magnetometer has been locked by the strong magnetic field of the high-voltage coil or is outputting a distorted signal. At this time, the system immediately generates a timing deviation flag. ; S14. One-way activation and joint parameter encapsulation of the irreversible drop verification state machine. The logic controller within the heterogeneous data processing module performs a strict Boolean AND operation on each of the input flag bits; the system simultaneously verifies the drop risk flag bit. Timing deviation flag If and only if the judgment condition is met, i.e. and This indicates that the system is currently in a dual high-risk state of "drop exceeding compensation limits" combined with "sensor physical timing collapse caused by a strong magnetic field"; subsequently, the system's main control microprocessor triggers a non-maskable interrupt, generating and fixing a one-way wake-up flag. Setting this flag to 1 will directly lock the drone's regular line-following autopilot task bus, forcing computing resources to switch to the highest priority "irreversible drop verification state machine"; to ensure the atomicity of data flow in cross-thread processing, the system will unidirectionally wake up the flag. Together with the background magnetic field gradient reference value retained at the front end Initial attitude transformation matrix They are packaged into a single high-priority state data structure and pushed onto the data bus.

[0021] In the preferred embodiment, step S2 includes the following steps: S21. Strong magnetic field gradient abrupt change monitoring and confidence level forced zeroing The heterogeneous data processing module detected a one-way wake-up flag. Immediately afterwards, the raw data of the real-time ambient magnetic field from the magnetometer were processed. Perform spatial differentiation to calculate the real-time magnetic field gradient at the current moment. The system will display the real-time magnetic field gradient. Reference value of background magnetic field gradient Perform differential operations to obtain the gradient drift. When the system recognizes At that moment, it was determined that the drone had entered a field of strong electromagnetic interference caused by high-voltage transmission towers, etc.; at this time, the system immediately triggered physical-level confidence allocation, forcibly incorporating the magnetometer fusion weight coefficients in the multi-source sensing Kalman filter model. Rewrite and lock to 0, physically cutting off the path for distorted magnetic field data to participate in attitude calculation; S22, Physical short-time integral takeover during the safety redundancy inertial period The system identifies the magnetometer fusion weighting coefficient. Then, the system's internal hardware timer automatically starts, timing one safe redundancy inertial cycle. The settings are in the hundreds of milliseconds range, covering the brief time it takes for a drone to pass through a high-voltage power line tower; Within the window, the system disables the magnetic field heading reference and forces entry into physical takeover mode; the system uses the initial attitude conversion matrix. Starting from the calculation point, extract real-time raw data from the triaxial accelerometer. The gravity vector in the data is used as a horizontal plane reference for correction, combined with the raw angular velocity data from the gyroscope in real time. Perform short-time continuous integration of quaternion differential equations; Specifically, the real-time attitude transformation matrix The integral update equation is as follows: (2); in, To calculate the generated real-time attitude transformation matrix; This is the initial attitude transformation matrix before the interference occurs; This is the attitude differential coupling function that combines the gravity vector and angular velocity; The time variable is the integral time variable, and its time domain is strictly limited to the safe redundancy inertial period. Inside; S23. Smooth recovery of background field alignment and collaborative fusion weights During the safe redundancy inertial period During or after the timing period, the system maintains a real-time monitoring of the magnetic field gradient. High-frequency monitoring; the system continuously calculates the real-time magnetic field gradient. Reference value of background magnetic field gradient The absolute difference between them; when detected After continuously fulfilling the set number of confirmation frames, the system determines that the drone has left the area of ​​strong electromagnetic interference and the environmental magnetic field divergence has returned to the background level; subsequently, the system releases the forced physical isolation state and integrates the magnetometer with the weighting coefficients. The system's default calibration parameters are reassigned from 0 to restore the standard collaborative pose fusion of the three-axis sensing hardware.

[0022] In the preferred embodiment, step S3 includes the following steps: S31. Extract the kinetic energy of physical destruction and read the instantaneous thrust limit of the motor. The heterogeneous data processing module receives the real-time attitude transformation matrix transmitted in step S2. Then, combined with the current raw data of the gyroscope's real-time angular velocity. Real-time raw data from triaxial accelerometers By using a rigid body dynamics model, the component of the drone's downward acceleration in the direction of gravity at the current moment is extracted and calculated. Yaw angular momentum The system will include the downward acceleration component. With the overall quality of the drone Multiply to obtain the destructive gravity that must be overcome vertically downwards; simultaneously combine with yaw angular momentum. With boom torque parameters The system performs a division operation to convert the current roll into the anti-torque thrust required to suppress the current roll; it then vectorizes these two forces to calculate the current destructive thrust required to overcome in order to restore stable hovering. ; Specifically, existing destructive thrust requirements The dynamic derivation equation is as follows: (3); in, To calculate the existing destructive thrust requirements; To pre-enter the overall quality of the drone into the system; The calculated components of the downward acceleration; This represents the current yaw momentum of the aircraft. These are the boom torque parameters; After completing the calculations, the system reads the maximum instantaneous full-load thrust limit parameters of the motors, which are permanently configured in the flight control electronic speed controller (ESC), via the underlying CAN bus. This parameter characterizes the maximum physical resultant force limit of the pull force that all the drive motors at the bottom layer of the system can provide under short-term 100% duty cycle full throttle pulse width modulation (PWM) output; S32. Dynamic Calculation of Thrust Compensation Deficit and Irreversible Anomaly Locking The heterogeneous data processing module obtains the existing destructive thrust requirements. With the motor's highest instantaneous full-load thrust limit parameter Then, the thrust compensation difference is calculated in real time. ; Specifically, thrust compensation difference The calculation equation is as follows: (4); in, This is the thrust compensation difference; This refers to the maximum instantaneous full-load thrust limit parameter of the motor; The existing destructive thrust requirement is calculated for step S31; The system calculates the thrust compensation difference. Perform symbol and temporal feature analysis: if If the system determines that the drone's physical recovery limit is sufficient to cover and offset the current destructive kinetic energy, the drone is still capable of restoring a stable attitude using its own power. At this point, the system resets its internal timer state without triggering an anomaly. If an anomaly is detected... The system indicated a thrust deficit, meaning its power output was absolutely insufficient to reverse the aircraft's downward trajectory. To filter out the extremely short-term overload disturbance caused by the sudden strong gusts of wind generated by the micro-topography of the barren mountains, the system, upon first detecting... The transient timer for gusts is activated to record the duration of the thrust deficit. The system performs iterative judgments, and the thrust deficit state lasts for a specified period of time. Greater than or equal to the preset gust transient filtering window At that point, the system confirms that the posture is beyond repair, completely locking it into an irreversible abnormal state, and immediately generates an irreversible abnormal lock flag. ; S33. After thorough verification, the execution command is pre-dispatched. The system microprocessor detected an irreversible exception lock flag in the interrupt service routine. Then, immediately activate the underlying lidar rangefinder or high-precision barometer in parallel to sample and obtain the drone's current real-time absolute altitude above the ground. The system synchronously reads the minimum safety distance for parachute deployment, which is pre-written into the flash memory configuration area. This parameter represents the minimum physical drop height difference required for the parachute system to fully inflate and deploy after ejection and establish effective air resistance; the system provides real-time absolute altitude above the ground. Minimum safety distance from umbrella opening Perform comparator operations; if This means the drone's descent trajectory has reached or even fallen below the minimum threshold required for safe parachute deployment, necessitating immediate intervention; or, if the drone is falling from an extremely high altitude, but its descent trajectory predicts... It is about to fall below within several machine cycles. Upon receiving the signal, the system immediately generates and sends a high-level irreversible drop confirmation signal through the hardware I / O port pin. This interrupts and freezes the normal command delivery path of the flight control module.

[0023] In the preferred embodiment, step S4 includes the following steps: S41, Triggering the underlying anti-tangling interlock mechanism and forcibly blocking power. Once the system master control unit or the underlying security coprocessor captures a high-level irreversible fall acknowledgment signal in its highest priority interrupt service routine (ISR), This immediately triggers the underlying anti-tangling interlock mechanism; to ensure that the execution of the power-off command is not affected by any upper-level crashes or software glitches, the system synchronously performs blocking operations on both the physical and logical links: On the logic link, the system directly pulls low the enable pin level of the communication control bus between the flight control board and all electronic speed controllers (ESCs), thus deactivating the real-time motor drive duty cycle signals of all rotor motors. Forced zeroing converts the signal to a zero-width truncation signal. ; On the physical link, for power inspection drones equipped with hardware safety redundancy, the system synchronously sends a circuit-breaking command to the normally closed solid-state relay coil located in the main power supply circuit to cut off the high-voltage power supply to the ESC board. After this joint action is completed, the system collects the feedback back electromotive force of each motor to confirm that the power output has been completely cut off, and then generates a power cut-off confirmation flag. ; S42. Mechanical deceleration dead zone redundancy calculation The system microprocessor detects the power cut-off confirmation flag. Immediately afterwards, the high-precision independent timer inside the microcontroller, unaffected by the main program, is activated; the system will then use the pre-calibrated mechanical deceleration dead time. The countdown initial value is loaded into the timer register; The mechanical deceleration dead time The specific values ​​range from 200ms to 400ms. Based on the rotational inertia parameters of the rotor assembly and the air resistance attenuation characteristics, they were pre-calculated through wind tunnel stop resistance experiments. During the countdown of the independent timer, the system locks the control bus of the slow descent actuator, prohibiting the issuance of any form of action command, so that the UAV is in a mechanical dead zone waiting state of pure gravity descent and rotor braking by air resistance. The system continuously monitors the remaining time of the independent timer. If and only if When the speed decreases to zero, the system, based on a preset physical model, confirms that due to the loss of electromagnetic drive torque, the propeller's residual rotational kinetic energy has undergone substantial physical attenuation under the action of aerodynamic drag, and the rotational speed has dropped below the safe threshold that will not cut the parachute lines; at this point, the system immediately generates a dead zone exhaustion confirmation flag. ; S43. End-of-line safety execution after mechanical deceleration dead zone exhaustion The descent control actuator receives a high-level dead-time exhaustion acknowledgment flag. Afterwards, the system logic confirms that the current state has flawlessly met all the physical prerequisites of "irreversible high-altitude crash" and "rotor power has completely decayed to the safe range"; Subsequently, the drive unit closes the high-current MOSFET circuit at its end, according to the igniter / solenoid valve trigger level threshold parameters. The system outputs a series of pulse-width action trigger commands to the igniter inside the parachute ejection tube or the solenoid valve of the high-pressure gas cylinder in the airbag. ; While this operation is being performed, the system's background data logging thread is activated, immediately obtaining the three-dimensional spatial coordinates of the crash location at the moment of impact via the satellite positioning module. It then invokes the non-volatile flash memory interface to transfer the aforementioned coordinate data and the real-time absolute altitude above the ground that led to the decision taking effect. All historical flags that trigger the entire process are packaged and written into the write-protected sector of the black box.

[0024] In a preferred embodiment, based on the same inventive concept as the aforementioned method embodiments, this application provides an active descent control system for unmanned aerial vehicles (UAVs) based on dynamic thrust deficit, specifically including: The multi-source timing wake-up unit employs attitude and heading reference hardware integrating an MPU-9250 nine-axis microelectromechanical system sensor. This unit acquires multi-source raw sensing data and initializes reference values; continuously compares the Z-axis acceleration drop component with the maximum redundant lift compensation limit; and outputs a unidirectional wake-up flag when a timing discrepancy is detected between the angular velocity and the rate of change of the magnetic field heading. ; The physical stripping takeover engine uses an STMicroelectronics STM32H7 series double-precision floating-point microcontroller; this engine is used to detect a one-way wake-up flag. Enable is triggered when set to 1; when a real-time magnetic field gradient drift exceeds the limit, the magnetometer fusion weighting coefficient is forcibly applied. Set to zero; then during the safe redundancy inertial period Internally, it performs rigid body quaternion differential and integral operations, and outputs a real-time attitude transformation matrix. ; The thrust deficit determination module uses an NXP i.MX8M Plus multi-core processor bridged to an IWR6843 millimeter-wave ranging radar; this module is used to analyze the real-time attitude transformation matrix. Calculate the existing destructive thrust requirements Compare this with the motor's highest instantaneous full-load thrust limit parameter. Calculate the thrust compensation difference by subtraction. When the difference is determined to be continuously less than zero and the real-time absolute altitude above the ground... When the price falls below the safety margin, an irreversible drop confirmation signal is issued. ; The timing interlock execution terminal employs an automotive-grade MOSFET power drive array combined with an independent hardware watchdog; this terminal is used to detect irreversible drop confirmation signals. At this time, the communication bus is disconnected, enabling the power supply to the electronic speed controller; the timer is activated to execute the mechanical deceleration dead time. Countdown; confirmation flags exhausted in dead zone After setting it to 1, output the action trigger command. And save the faulty snapshot to disk; Furthermore, based on the same inventive concept as the aforementioned method embodiments, this application embodiment also provides a computer-readable storage medium. This computer-readable storage medium uses an industrial wide-temperature range NAND non-volatile flash memory physical chip, which stores computer-executable instructions. When the computer-executable instructions are read and executed, all steps of the above-mentioned UAV active descent control method based on dynamic thrust deficit are implemented.

[0025] The above embodiments are merely preferred technical solutions of the present invention and should not be regarded as limitations on the present invention. The scope of protection of the present invention should be the technical solution described in the claims, including equivalent substitutions of the technical features described in the claims; that is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for active descent control of unmanned aerial vehicles based on dynamic thrust deficit, characterized in that, Includes the following steps: S1. Obtain raw angular velocity data Raw magnetic field data Compared with raw acceleration data Calculate and generate background magnetic field gradient reference values With the initial attitude transformation matrix In the raw acceleration data Exceeding the limit, and based on raw angular velocity data With the original magnetic field data Calculated heterogeneous data deviation coefficient When the limit is exceeded, output a one-way wake-up flag. ; S2, Obtain the one-way wake-up flag. Then, based on the real-time magnetic field gradient Reference value of background magnetic field gradient The difference will be used to integrate the magnetometer weighting coefficient. Set to zero; during the safe redundancy inertial period Internally, based on the initial attitude transformation matrix Angular velocity raw data Compared with raw acceleration data Perform continuous integration to generate a real-time attitude transformation matrix. ; S3, Based on real-time attitude transformation matrix Perform dynamics calculations to generate the existing failure thrust requirements. Based on the motor's maximum instantaneous full-load thrust limit parameters Compared with existing destructive thrust requirements The thrust difference and the real-time absolute altitude above the ground Perform a graded judgment and output an irreversible fall confirmation signal when the irreversible fall condition is met. ; S4. Receive irreversible drop confirmation signal Then, a cutoff signal was sent. And disconnect the main power supply circuit to generate a power disconnection confirmation flag. ; Confirmation of power cut-off flag Startup mechanical deceleration dead time A countdown timer will run out, after which an action trigger command will be output. Drive the descent actuator and store the fault snapshot dataset. .

2. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S1 specifically includes: Extract raw acceleration data Instantaneous value of vertical axis acceleration in ; Instantaneous value of vertical axis acceleration Maximum Redundancy Lift Compensation Limit Conduct comparative evaluation; Determining the instantaneous value of vertical axis acceleration Breaking the maximum redundancy lift compensation limit When this happens, an internal timer is triggered to open the initial monitoring time window. and in the initial monitoring time window Internal activation heterogeneous sensor timing coupling deviation verification mechanism.

3. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S1 specifically includes: Extract raw angular velocity data separately Dynamic mutation rate and original magnetic field data The rate of change of course; Combined with static proportional mapping coefficients Dimensional alignment was performed on the dynamic mutation rate and the rate of yaw change to assess the physical coupling consistency between the dynamic mutation rate and the rate of yaw change in the kinematics of the rigid body in space, and the heterogeneous data bias coefficient was determined. ; heterogeneous data deviation coefficient Physical coupling limit threshold A comparison was performed to determine the deviation coefficient of the heterogeneous data. Exceeding the physical coupling limit threshold At that time, a one-way wake-up flag is generated. Forced entry into the irreversible drop verification state machine.

4. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S2 specifically includes: Comparison of real-time magnetic field gradients Reference value of background magnetic field gradient The gradient drift is acquired, and when the gradient drift exceeds a preset step threshold, a physical-level confidence assignment is triggered to lock the magnetometer fusion weight coefficients in the multi-source sensing fusion model. Zero; During the safe redundancy inertial period of starting the timing. Inside, shield the magnetic field heading reference data and use the initial attitude transformation matrix. Starting from the calculation point, combined with the original acceleration data Extracted raw data of gravity vector and angular velocity Perform continuous integration of the rigid body space differential equations to generate a real-time attitude transformation matrix. .

5. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S3 specifically includes: Based on real-time attitude transformation matrix Angular velocity raw data and raw acceleration data Extracting the downward acceleration component Yaw angular momentum ; The extracted fall acceleration components are based on the overall mass of the drone. Mapped to the initial thrust required to overcome the downward fall, and the yaw momentum of the airframe combined with the arm moment parameters. This is converted into a second thrust requirement to suppress the body's rollover. Vector synthesis is performed on the first thrust requirement and the second thrust requirement to generate the existing destructive thrust requirement. .

6. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S3 specifically includes: Compare the maximum instantaneous full-load thrust limit parameters of the motor Compared with existing destructive thrust requirements Generate thrust compensation difference that characterizes the dynamic supply and demand boundary. ; thrust compensation difference It is in a negative state and continues to exceed the transient gust filter window. When the attitude is determined to be in an irrecoverable state, the ranging sensing hardware is activated in parallel to read the real-time absolute altitude above the ground. ; Real-time absolute altitude above the ground Minimum safety distance from umbrella opening Comparison, to confirm the real-time absolute altitude above the ground Falling below the minimum safety distance for parachute opening When this occurs, an irreversible drop confirmation signal is output in a high-level state. .

7. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S4 specifically includes: Capture irreversible fall confirmation signal Then, an anti-entanglement interlock mechanism is triggered synchronously on both the logical and physical links; Pull the communication control bus enable pin low on the logic link to send a zero-width truncation signal. The motor drive duty cycle is reset to zero; a circuit breaker command is sent to the main power supply circuit in parallel on the physical link to cut off the high power supply to the electronic speed controller; After confirming that the logical link and physical link have been successfully blocked and the underlying power output has been cut off, a power cut-off confirmation flag is generated. .

8. The active descent control method for unmanned aerial vehicles based on dynamic thrust deficit according to claim 1, characterized in that, Step S4 specifically includes: Based on the power cut-off confirmation flag The pre-calibrated mechanical deceleration dead time An independent hardware timer is loaded to execute the countdown, and the control bus of the slow-descent actuator is locked to wait for the rotor's mechanical kinetic energy to cause air resistance decay; Dead time of mechanical deceleration When the countdown expires, confirm the dead zone exhaustion flag. When in an active state, the closed-end drive unit circuit outputs an action trigger command. Activate the igniter or solenoid valve; Parallel acquisition of the 3D spatial coordinates of the crash location and state machine transition feature parameters at the moment of the crash, and packaging them to construct a fault snapshot dataset. It is then written to the external flash memory black box.

9. An active descent control system for unmanned aerial vehicles (UAVs) based on dynamic thrust deficit, characterized in that, include: The multi-source timing wake-up unit is used to acquire raw sensing data of angular velocity, magnetic field, and acceleration, and to calculate the background magnetic field gradient reference value. With the initial attitude transformation matrix ; and the deviation coefficient between acceleration components and heterogeneous data When both limits are exceeded, output a one-way wake-up flag. ; Physical stripping takeover engine for responding to one-way wake-up flags The magnetometer is integrated with weighting coefficients based on the magnetic field gradient difference. Set to zero; In the safe redundancy inertial period Perform continuous integration operations to generate a real-time attitude transformation matrix. ; The thrust deficit determination module is used to determine the thrust deficit based on the real-time attitude transformation matrix. Solving for existing destructive thrust requirements ; Combining the motor's full-load thrust limit parameters Real-time absolute altitude above the ground Perform a drop test and output an irreversible drop confirmation signal. ; Timing interlock execution terminal, used to respond to irreversible drop confirmation signal. Send cutoff signal And disconnect the power supply circuit to generate a power disconnection confirmation flag. ; In mechanical deceleration dead time After the countdown expires, output the action trigger command. Drive the slow descent actuator and store the fault snapshot dataset. .

10. A computer-readable storage medium having internally stored computer-executable instructions, characterized in that: When the computer-executable instructions are read and executed by the processor, the active descent control method for unmanned aerial vehicles based on dynamic thrust deficit, as described in any one of claims 1 to 8, is implemented.

Images