Unmanned aerial vehicle centimeter-level hovering control system based on single Beidou differential positioning

Abstract

The invention discloses an unmanned aerial vehicle centimeter-level hovering control system based on single Beidou differential positioning, and particularly relates to the technical field of unmanned aerial vehicle high-precision positioning and control. The system comprises a Beidou RTK positioning module, a multi-sensor fusion module, a flight control module, a motor driving module, an exception handling module and a ground base station module. Centimeter-level positioning data are obtained through Beidou RTK, and a high-precision pose is output through tight coupling filtering by combining an IMU and a barometer; the flight control module adopts a model prediction algorithm to generate an instruction; the motor driving module realizes quick response; the exception handling module guarantees the reliability of the system; the ground base station provides differential correction. The single Beidou differential positioning and multi-sensor fusion technology is innovatively adopted, centimeter-level precision autonomous hovering control of the unmanned aerial vehicle is achieved, and the unmanned aerial vehicle hovering control system has the advantages of being high in precision, high in interference resistance and rapid in response and is suitable for industrial unmanned aerial vehicle application scenes such as precision agriculture and electric power inspection.

Description

Technical Field

[0001] The present invention discloses a centimeter-level hovering control system for unmanned aerial vehicles (UAVs) based on single Beidou differential positioning, and particularly relates to the technical field of high-precision positioning and control of UAVs. Background Art

[0002] Existing high-precision drone positioning and control systems still face numerous technical bottlenecks in practical applications, severely restricting their effectiveness in industrial scenarios. Regarding satellite positioning, while RTK technology theoretically offers centimeter-level positioning accuracy, existing solutions generally rely on the combined computation of multiple satellite systems. This leads to a sharp decline in performance in scenarios such as urban canyons and near-Earth flight, where signals from a single satellite system are limited. Furthermore, conventional RTK technology is sensitive to the distance from the base station, and the problem of rapid degradation of positioning accuracy with increasing distance has remained unresolved.

[0003] In the field of sensor fusion, the currently mainstream loosely coupled architecture suffers from significant technical flaws. This architecture simply combines the outputs of individual sensors, failing to fully exploit the complementary nature of multi-source data. Specifically, the lack of an online calibration mechanism for the IMU's angular velocity integral error leads to drift in pose estimation over time. The lack of an effective strategy for compensating for barometer environmental errors makes it difficult to guarantee altitude measurement accuracy. More importantly, when satellite signals are interrupted, it is difficult to provide an accurate calibration baseline for inertial navigation, severely impacting system reliability.

[0004] Flight control systems also face significant challenges. Traditional PID control algorithms have three inherent flaws: fixed control parameters make it difficult to adapt to the dynamic characteristics of drones under varying flight conditions; their ability to compensate for system lags is limited, resulting in delayed control response; and their poor anti-interference capabilities make them prone to sustained oscillations under sudden wind disturbances. These issues are particularly prominent in applications requiring centimeter-level hovering accuracy.

[0005] In addition, the shortcomings of existing technologies in terms of system fault tolerance also restrict their practical application value. On the one hand, the lack of intelligent sensor fault detection mechanism makes it difficult to detect and isolate abnormal data in a timely manner; On the other hand, the degraded operation strategy is overly conservative, often triggering a return or landing in abnormal situations such as satellite signal loss, seriously affecting operational continuity. These technical shortcomings make the existing system unable to meet the stringent high-precision drone operation requirements in professional fields such as power inspection and precision agriculture.

[0006] Therefore, in response to the many shortcomings of existing high-precision positioning and control technologies for drones, this paper proposes an innovative centimeter-level drone hovering control system based on single Beidou differential positioning. By adopting single Beidou RTK high-precision positioning technology, this system breaks through the traditional solution's reliance on multi-satellite systems and significantly improves the system's usability in complex environments. The innovative tightly coupled filtering architecture fully exploits the complementary characteristics of multi-source sensor data, effectively solving key issues such as IMU drift and barometer error compensation. The advanced model predictive control algorithm achieves accurate modeling and rapid response to the drone's dynamic characteristics, significantly improving its anti-interference capability. At the same time, the intelligent exception handling mechanism ensures the system's reliable operation under various abnormal conditions. Summary of the Invention

[0007] The purpose of this invention is to provide a centimeter-level hover control system for unmanned aerial vehicles (UAVs) based on single Beidou differential positioning. This system uses a Beidou RTK high-precision positioning module to obtain centimeter-level position coordinates, and employs a tightly coupled filtering algorithm to deeply fuse Beidou positioning data with information from multiple sensors, such as the IMU and barometer, to output a high-precision pose estimate. The flight control module generates optimized control instructions based on a model predictive control algorithm, and achieves precise execution through a high-response motor drive module. The system is also equipped with an intelligent exception handling module and a ground reference station differential service, which together ensure stable and reliable centimeter-level hover control for UAVs in complex environments. This invention breaks through the traditional solution's reliance on multiple satellite systems, significantly improving positioning accuracy, environmental adaptability, and control stability.

[0008] The purpose of the present invention can be achieved through the following technical solutions: A centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning, characterized in that the system includes a Beidou RTK high-precision positioning module, a multi-sensor data fusion module, a flight control module, a motor drive and execution module, an exception handling and fault tolerance module, and a ground reference station module, wherein: The BeiDou RTK high-precision positioning module is used to perform real-time dynamic carrier phase differential processing on the UAV positioning signal based on the original BeiDou satellite observation data and the differential correction number provided by the ground reference station, thereby generating real-time position coordinates with centimeter-level accuracy; The multi-sensor data fusion module is used to use the tightly coupled filtering algorithm and the centimeter-level position data provided by the Beidou RTK module to perform multi-source data fusion on the IMU raw measurement values ​​and barometric altitude information, and ultimately output a high-reliability pose estimate with high-frequency updates; The flight control module is used to implement rolling time domain optimization control of the UAV motion state using the model predictive control algorithm and the posture data provided by the multi-sensor fusion module to generate the motor control quantity within the cycle; The motor drive and execution module is used to precisely drive the brushless motor through PWM modulation technology and control instructions issued by the flight control module to achieve millisecond-level response power output; The exception handling and fault tolerance module is used to diagnose and grade system anomalies using a multi-dimensional health monitoring algorithm and real-time operation data of each module, and output downgrade operation instructions or emergency protection signals; The ground reference station module is used to perform real-time differential calculations on satellite observation errors based on a high-precision Beidou reference receiver and known reference station coordinates, and continuously broadcast differential correction numbers with millimeter-level accuracy through a 4G communication link.

[0009] Optionally, when the BeiDou RTK high-precision positioning module performs filtering optimization on pseudorange observation noise based on carrier phase smoothing technology, the method includes: Relying on the carrier phase continuity feature, the pseudo-range observation values ​​are processed by sliding average to obtain a smooth observation sequence; Combined with the satellite elevation angle parameters, the filtering weights are dynamically allocated to form the optimal weighted result; The cycle slip detection algorithm is used to automatically remove abnormal data to ensure observation continuity.

[0010] Optionally, when the BeiDou RTK high-precision positioning execution module performs real-time correction on the satellite orbit error based on the reference station differential correction number, it includes: The dual-frequency observation combination is used to separate the ionospheric delay error and generate the ionospheric correction. With the help of meteorological observation data, the tropospheric refraction is compensated by the model and the atmospheric correction parameters are output; The satellite orbit deviation is differentially eliminated using the precise coordinates of the reference station to generate a position correction result.

[0011] Optionally, when the multi-sensor data fusion module performs alignment processing on heterogeneous sensor data based on a time synchronization mechanism, the multi-sensor data fusion module includes: Use hardware trigger signals to synchronize multi-source sensor data and establish a unified time base; Relying on the interpolation compensation algorithm, the asynchronous sampling data is time-aligned to form a synchronous observation sequence.

[0012] Optionally, when the multi-sensor data fusion module performs optimal weighting on the pose estimation error based on the tightly coupled filtering algorithm, the multi-sensor data fusion module includes: Adaptive Kalman filtering is used to complete the covariance estimation of multi-source observation errors and construct the optimal weight matrix; Combined with the residual detection mechanism, abnormal observation data can be automatically eliminated to ensure the reliability of the fusion results.

[0013] Optionally, when the flight control module performs the prediction of the six-degree-of-freedom motion completion state of the UAV based on dynamic modeling, the flight control module includes: The rigid body dynamics model is used to predict the UAV's motion state and generate a state estimate. The aerodynamic parameter library is used to complete the modeling and compensation of wind disturbance effects and output anti-disturbance prediction results.

[0014] Optionally, when the flight control module performs closed-loop adjustment on the posture tracking deviation based on the error feedback mechanism, the flight control module includes: Relying on proportional-integral-differential control, the posture deviation is gradually eliminated to generate a steady-state control variable; The feedforward compensation technology is used to dynamically correct the system lag effect and form a fast response instruction.

[0015] Optionally, when the motor driving and executing module executes power conversion based on the PWM modulation technology for the control instruction, it includes: Use space vector modulation to synthesize the waveform of three-phase voltage instructions and generate drive pulse signals; The dead-time compensation technology is used to minimize switching losses and ensure power conversion efficiency.

[0016] Optionally, when performing real-time monitoring of sensor data quality based on health assessment, the exception handling and fault tolerance module includes: Use the chi-square test method to quickly detect abnormal sensor data and trigger a fault alarm; Relying on the voting decision-making mechanism, the redundant system can achieve mode switching and maintain basic operating functions.

[0017] Optionally, when the ground reference station module performs modeling compensation for regional atmospheric errors based on multi-base station joint solution, it includes: Use network RTK technology to complete spatial modeling of regional atmospheric errors and construct error correction fields; The differential correction data can be broadcast in real time through the data broadcast protocol to provide positioning enhancement services.

[0018] This invention provides a centimeter-level hover control system for unmanned aerial vehicles (UAVs) based on single Beidou differential positioning. This system innovatively utilizes single Beidou RTK high-precision positioning technology. Through carrier phase differential processing and a pseudorange smoothing algorithm, it achieves centimeter-level positioning accuracy solely based on the Beidou satellite system, overcoming the traditional solution's reliance on multiple satellite systems for joint positioning. The ground reference station module broadcasts differential corrections in real time via a 4G link, providing a precise positioning reference for the UAV and significantly improving the system's availability and reliability in complex environments.

[0019] The system's core innovation lies in its tightly coupled multi-sensor fusion architecture. This architecture precisely aligns BeiDou RTK positioning data, IMU inertial measurement data, and barometric altitude data through a time synchronization mechanism. It employs an adaptive Kalman filter algorithm for optimal weighted fusion, effectively mitigating IMU drift and barometer error accumulation. The fused pose estimate data is updated at a 100Hz frequency, providing high-precision and reliable state feedback for flight control, ensuring stable system operation in a wide range of environmental conditions.

[0020] At the control level, this invention employs a model predictive control algorithm, performing rolling-time optimization based on the drone's dynamics model, to generate control instructions with a 20ms period. The motor drive module utilizes 400Hz PWM modulation technology to achieve millisecond-level response. Combined with the multi-level fault-tolerance mechanism of the exception handling module, this allows the drone to maintain centimeter-level hovering accuracy even in the event of brief satellite signal interruptions or sensor anomalies.

[0021] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention innovatively utilizes single-Beidou RTK carrier phase differential technology, combined with advanced pseudorange smoothing algorithms and multipath error suppression methods, to achieve high-precision positioning solely based on the Beidou satellite system. Compared to traditional multi-system fusion solutions, this technology not only simplifies the system architecture but also significantly improves positioning stability in complex environments. In particular, in scenarios where satellite signals are limited, such as urban canyons and near-Earth flight, the system maintains excellent positioning performance, providing drones with a continuous and reliable centimeter-level position reference.

[0022] 2. The system's high-frequency, tightly coupled fusion architecture, combined with a model predictive control algorithm, enables precise control of drones in dynamic environments. This innovative control structure effectively compensates for various delays and nonlinearities, ensuring exceptional stability in the face of external disturbances such as wind disturbances. This control scheme is particularly well-suited for operations requiring precise positioning, significantly enhancing the drone's operational capabilities and safety in complex conditions.

[0023] 3. Through its intelligent, multi-level fault-tolerance design, the system possesses exceptional anomaly-response capabilities. Its innovative fault diagnosis algorithm monitors the status of each sensor in real time and automatically selects the optimal operating mode. Even in the event of a temporary satellite signal interruption or sensor anomaly, the system maintains reliable positioning and control performance, significantly extending operational time in these conditions and significantly improving mission completion rates and system reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a structural schematic diagram of a centimeter-level hovering control system for an unmanned aerial vehicle based on single Beidou differential positioning according to the present invention. DETAILED DESCRIPTION

[0025] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] Example: Figure 1 The present invention provides a structural schematic diagram of a centimeter-level hovering control system for a UAV based on single Beidou differential positioning. The centimeter-level hovering control system for a UAV based on single Beidou differential positioning includes the following modules: BeiDou RTK high-precision positioning module: Based on the original BeiDou satellite observation data and the differential corrections provided by the ground reference station, it performs real-time dynamic carrier phase differential processing on the UAV positioning signal, thereby generating real-time position coordinates with centimeter-level accuracy; Multi-sensor data fusion module: This module uses a tightly coupled filtering algorithm and centimeter-level position data provided by the BeiDou RTK module to fuse IMU raw measurements and barometric altitude information, ultimately outputting a highly reliable pose estimate with high-frequency updates. Flight control module: uses the model predictive control algorithm and the posture data provided by the multi-sensor fusion module to implement rolling time domain optimization control of the UAV's motion state and generate the motor control quantity within the cycle; Motor drive and execution module: Through PWM modulation technology and control instructions issued by the flight control module, the brushless motor is precisely driven to achieve millisecond-level response power output; Abnormal handling and fault tolerance module: Utilizes multi-dimensional health monitoring algorithms and real-time operation data of each module to diagnose and classify system abnormalities, and output degraded operation instructions or emergency protection signals; Ground reference station module: Based on the high-precision BeiDou reference receiver and known reference station coordinates, it performs real-time differential calculations on satellite observation errors and continuously broadcasts differential corrections with millimeter-level accuracy through the 4G communication link.

[0027] In an embodiment of the present invention, when the BeiDou RTK high-precision positioning module performs filtering optimization on pseudo-range observation noise based on carrier phase smoothing technology, the following processing flow is specifically included: The system deeply analyzes the continuous nature of carrier phase observations and constructs a sliding window-based digital filtering model. This model uses a recursive least squares algorithm to optimally estimate the pseudorange observation sequence within the window, effectively suppressing the influence of random noise. The system also features a dynamic window adjustment mechanism that intelligently adjusts the time span of the smoothing window based on the satellite signal-to-noise ratio and carrier phase lock status, ensuring both smoothing and real-time performance.

[0028] The system establishes an elevation-weight mapping model for satellite signals at different elevation angles: high-elevation-angle satellites (which typically have better signal quality) are assigned a larger smoothing weight, while low-elevation-angle satellites (which are more susceptible to multipath) have their weights appropriately reduced. This differentiated processing significantly improves the system's observation quality in complex environments, especially in scenarios with severe multipath effects, such as urban canyons.

[0029] The system integrates an advanced real-time cycle slip detection and repair mechanism. By analyzing the second-order difference characteristics of carrier phase observations and combining them with consistency checks on multi-band observation data, a highly sensitive cycle slip detection algorithm has been constructed. This algorithm not only accurately identifies the moment of cycle slip occurrence but also distinguishes between cycle slips and phase changes caused by normal dynamics. Once a cycle slip event is detected, the system immediately initiates a smoothing reset procedure and rapidly recovers the full-cycle ambiguity using wide-lane combined observations.

[0030] By monitoring the residual sequence of observations before and after smoothing in real time, the smoothing effect is evaluated and algorithm parameters are dynamically adjusted. The system also records historical smoothing data for each satellite and establishes a satellite signal quality assessment database, providing a reference for subsequent observation processing. This quality control mechanism ensures the reliability and stability of the smoothing process.

[0031] In an embodiment of the present invention, when the BeiDou RTK high-precision positioning module performs real-time correction of satellite orbit errors based on the reference station differential correction number, the processing flow specifically includes the following: The system leverages the ionospheric dispersion characteristics of the BeiDou dual-frequency signal to establish an ionospheric delay estimation model based on geometrically inert phase combinations. This model uses a sliding window least squares algorithm to calculate ionospheric delay parameters in real time. Inter-frequency bias correction effectively eliminates the effects of first-order ionospheric terms, significantly improving the usability of low-altitude satellite observation data.

[0032] To address tropospheric delay errors, the system constructed an improved compensation model that integrates measured meteorological data. Temperature, pressure, and humidity observations acquired through the base station network are used to calculate the static delay component using the Saastamoinen model, and the wet delay component is processed using a global mapping function. Specifically tailored to the characteristics of low-altitude UAV operations, the delay modeling algorithm for the near-ground atmosphere has been optimized, ensuring high positioning accuracy for satellite observations below an altitude angle of 15 degrees.

[0033] The system continuously monitors the deviation between the known reference station coordinates and the coordinates calculated from the satellite ephemeris. Through parametric modeling, it decomposes the orbit error into three orthogonal components: radial, along-track, and normal. Based on observational data from a regional reference station network, a weighted least squares algorithm is used to estimate orbit error correction parameters in real time, generating dynamic corrections applicable to different satellites and time periods.

[0034] The system utilizes a dedicated data compression algorithm and encapsulation format to integrate ionospheric, tropospheric, and orbital corrections into compact data packets. A low-latency, highly reliable data transmission link is established over the 4G LTE network, with forward error correction coding and automatic retransmission mechanisms ensuring the integrity of correction data transmission. The system also implements intelligent network quality monitoring, dynamically adjusting data transmission rates and priorities based on channel conditions.

[0035] In an embodiment of the present invention, when the multi-sensor data fusion module performs alignment processing on heterogeneous sensor data based on a time synchronization mechanism, the process specifically includes the following steps: The system incorporates a dedicated synchronization signal generator, generating a highly stable pulse signal as a global time reference. This signal is transmitted directly to each sensor node via hardware circuitry. The rising edge accuracy of the synchronization pulse reaches nanoseconds, providing a unified clock reference for heterogeneous sensors such as IMUs, BeiDou receivers, and barometers.

[0036] To address the issue of sensor sampling frequency discrepancies, the system implements resampling based on a fifth-order Lagrangian interpolation algorithm. This algorithm employs a sliding window strategy, dynamically adjusting the interpolation order based on sensor characteristics to achieve uniform data rates while preserving signal characteristics. High-frequency IMU data is downsampled, while low-frequency pressure data is upsampled, ensuring consistent temporal resolution across all sensor outputs.

[0037] The system incorporates a comprehensive sensor delay calibration system. A dedicated test platform measures each sensor's end-to-end delay, from signal acquisition to data output, including analog-to-digital conversion time, data processing time, and communication transmission delay. These calibration parameters are stored in non-volatile memory and automatically loaded upon system power-up for real-time compensation. The calibration process considers the effects of temperature variations and establishes a temperature-delay compensation model.

[0038] The system accurately time-stamps data at both the data acquisition and fusion processing ends, calculating actual transmission delays through closed-loop verification. An adaptive threshold detection algorithm is designed to provide real-time alerts and compensation for abnormal time deviations. A synchronized status monitoring interface intuitively displays the time deviations of each sensor, facilitating system debugging and maintenance.

[0039] The resulting time synchronization performance meets the stringent requirements of fusion processing. The time alignment error of all sensor data is controlled within the tolerances of the fusion algorithm, ensuring that tightly coupled filtering introduces no additional timing errors. The system supports automatic time resynchronization for hot-swappable sensors, ensuring rapid restoration of synchronization after sensor replacement or reboot.

[0040] In an embodiment of the present invention, when the multi-sensor data fusion module performs optimal weighting of the pose estimation error based on the tightly coupled filtering algorithm, the processing flow specifically includes the following: The system features a filtering framework with multiple adaptive mechanisms. This framework uses sliding window covariance estimation techniques to calculate the statistical characteristics of each sensor's observation noise in real time. This framework includes a dynamic noise covariance matrix adjustment algorithm that automatically optimizes the filter parameter configuration based on factors such as satellite geometry, signal environment, and sensor operating status.

[0041] The system constructs a multidimensional residual detection space and identifies anomalous observations through statistical hypothesis testing. For detected outliers, the system dynamically adjusts their weight coefficients in the new information sequence and triggers the sensor health assessment process. It also retains historically valid observation data to ensure the continuity of the filtering process, regardless of any single anomaly.

[0042] By analyzing the angular velocity and acceleration characteristics of the IMU output, the system automatically identifies the drone's current motion state, including static, uniform velocity, and maneuvering. For each state, it presets the optimal set of observation model parameters, including configuration parameters for the process noise matrix, observation noise matrix, and state transition matrix. This adaptive mechanism significantly improves the robustness of the filtering algorithm in different motion scenarios.

[0043] The system not only outputs the optimal fused pose estimate but also provides comprehensive accuracy assessment data, including position DOP, attitude accuracy metrics, and covariance matrix. These metrics are transmitted to upper-level control modules via standard interfaces, providing a reliable reference for flight control decisions. The system also records historical fusion performance data for long-term sensor performance evaluation and algorithm optimization.

[0044] The system implements real-time filtering performance monitoring and optimization. Key metrics such as sensor contribution, residual distribution, and convergence status are displayed through a visual interface. An automatic parameter tuning algorithm is designed to continuously optimize filtering parameters based on long-term operational data. It also supports online algorithm switching, dynamically selecting between different algorithms such as the Extended Kalman Filter and the Unscented Kalman Filter, ensuring the most appropriate fusion strategy for the scenario is always used.

[0045] In an embodiment of the present invention, when the flight control module performs the prediction of the six-degree-of-freedom motion completion state of the UAV based on dynamic modeling, the processing flow specifically includes the following: This model, based on the Newton-Euler equations, fully accounts for both translational and rotational dynamics in the aircraft coordinate system. The system integrates a database of key aerodynamic parameters, including lift, drag, and torque coefficients, and uses parametric modeling to accurately describe the coupling between rotor aerodynamic characteristics and aircraft motion. The model specifically optimizes aerodynamic calculation accuracy in low-speed flight conditions, ensuring accurate predictions during hovering and low-speed maneuvers.

[0046] The system continuously updates model parameters using a recursive least squares algorithm, collecting real-time sensor data such as motor speed, battery voltage, and aircraft angular rate. A specialized motion excitation strategy is designed to proactively stimulate specific motion modes while ensuring flight safety, improving parameter identification accuracy. The calibration process considers environmental factors such as temperature and altitude, establishing a multidimensional parameter compensation table to ensure high model accuracy under various environmental conditions.

[0047] The system builds a wind disturbance model based on an extended state observer. By analyzing the deviation between the actual motion state and the expected state, it estimates the magnitude and direction of wind disturbances in real time. A layered compensation strategy is designed, decomposing the disturbance compensation into two components: feedforward compensation and feedback compensation, which are injected into different parts of the control loop. The system can also identify different types of wind disturbances, such as gusts and turbulence, and automatically adjust the compensation algorithm parameters.

[0048] The core dynamics model runs at 100Hz and outputs complete motion state information, including position, velocity, attitude angle, and angular rate. The system incorporates a prediction result credibility assessment algorithm that analyzes prediction residuals and covariance matrices to assess prediction accuracy in real time. Prediction results utilize a timestamp alignment mechanism to ensure temporal consistency with sensor observations, providing accurate state estimates for subsequent control algorithms.

[0049] In an embodiment of the present invention, when the flight control module performs closed-loop adjustment of the posture tracking deviation based on the error feedback mechanism, the process specifically includes the following steps: The system utilizes a finite state machine-based control mode management system to automatically select the optimal set of control parameters based on the drone's current flight phase, motion state, and environmental conditions. Specific PID parameter combinations are pre-set for different flight modes, such as hovering, cruising, and maneuvering. Fuzzy logic algorithms enable smooth transitions between modes, preventing oscillations caused by sudden changes in control parameters.

[0050] The feedforward channel utilizes a model-based inverse dynamics calculation method to generate the ideal control variable based on the desired trajectory, preemptively compensating for system inertia and actuator delay. The feedback channel incorporates an adaptive PID algorithm that automatically adjusts the integral and differential time parameters by analyzing the error rate in real time, effectively eliminating steady-state errors. The outputs of the two channels are optimally weighted and fused to ensure both fast dynamic response and stable tracking accuracy.

[0051] Independent PID controllers are designed for the three rotational degrees of freedom (Roll, Pitch, and Yaw), and frequency-domain analysis is used to optimize the control bandwidth of each axis. For altitude and horizontal position control, a cascade control structure based on acceleration feedforward is established. The system uses a dynamic coupling evaluation algorithm to monitor the interactions between the various degrees of freedom in real time, introducing decoupling compensation terms when necessary.

[0052] The system continuously monitors each motor's command value and actual output, predicting potential saturation risks through back-calculation. A conditional integration algorithm automatically freezes the integral term when signs of saturation are detected, preventing overshoot caused by integral saturation. A dynamic command limiting function is also implemented, adjusting the maximum allowable control value in real time based on battery voltage and motor temperature to ensure that the actuator always operates within the linear range.

[0053] In an embodiment of the present invention, when the motor drive and execution module performs power conversion based on the PWM modulation technology for the control instruction, the process specifically includes the following: The system converts a three-phase coordinate system into a two-phase stationary coordinate system using the Clarke transform and generates an optimal PWM switching sequence using a vector synthesis algorithm. Compared to traditional sinusoidal PWM technology, this modulation method improves DC bus voltage utilization by approximately 15%, significantly enhancing the motor's output capability under low-voltage conditions. The system calculates the voltage vector action time and switching sequence in real time to minimize switching losses in power devices.

[0054] To address the dead-zone effect during power tube switching, the system establishes an adaptive compensation model based on current polarity detection. A high-precision timer captures the rising and falling edges of the actual output pulse and dynamically adjusts the PWM signal's duty cycle compensation. The compensation algorithm considers parameters such as the power tube's conduction characteristics and reverse recovery time, achieving nanosecond-level timing accuracy and effectively eliminating output voltage waveform distortion.

[0055] The current closed-loop control utilizes a three-loop nested architecture to achieve precise torque regulation. The innermost loop is a Hall-effect sensor-based current sampling loop with a bandwidth designed to exceed 10kHz. The middle loop is a current decoupling loop based on field-oriented control, enabling independent control of the d-axis and q-axis currents. The outermost loop is the torque regulation loop, which converts the controller's torque command into a corresponding current reference value. The system integrates online parameter identification, automatically updating key parameters such as motor resistance and inductance, ensuring that the control algorithm always matches the actual motor characteristics.

[0056] Multiple protection mechanisms establish a comprehensive system safety assurance framework. Overcurrent protection utilizes dual detection using hardware comparators and software algorithms, with a response time of less than 2 microseconds. Overtemperature protection utilizes gradient monitoring via multiple temperature sensors distributed throughout the power module and motor windings. Undervoltage protection incorporates a hysteresis comparator circuit to prevent malfunctions caused by voltage fluctuations. All protection signals are transmitted to the main controller via optoelectronic isolation, ensuring rapid shutdown of the drive signal in the event of a fault. The system also implements self-diagnosis and records historical fault information to aid maintenance.

[0057] In an embodiment of the present invention, when the exception handling and fault tolerance module performs real-time monitoring of sensor data quality based on health assessment, the process specifically includes the following: The system builds a three-tiered evaluation framework encompassing data quality, device status, and system performance indicators. By quantitatively analyzing characteristic parameters such as continuity, integrity, and rationality of sensor data, a dynamic health scoring model is established. This model comprehensively considers historical data trends and real-time status changes, employing fuzzy logic algorithms to calculate the comprehensive health score for each module, enabling early warning of anomalies.

[0058] The system establishes a fault tree database encompassing typical failure modes, including sensor failures, communication anomalies, and actuator failures. Each terminal event is associated with multiple intermediate and base events. By matching system anomaly signatures with fault tree nodes in real time, the root cause of the fault can be quickly located. The diagnostic process incorporates a Bayesian inference algorithm, combining prior probabilities with real-time observational data to calculate the credibility of each fault hypothesis, improving the accuracy of complex fault diagnosis.

[0059] The system categorizes fault severity into four levels: minor, general, major, and fatal, each corresponding to different fault-tolerance measures. Minor faults trigger adaptive parameter adjustments; general faults initiate redundant module switching; major faults implement functional degradation; and fatal faults immediately enter safety protection mode. Fault-tolerance decisions utilize a voting mechanism that integrates the results of multiple health assessment indicators to make a final decision, avoiding unnecessary switching due to misjudgment of a single indicator.

[0060] The system records the complete handling process of all abnormal events, including data such as fault characteristics, diagnostic results, and treatment effectiveness. Machine learning algorithms analyze historical cases to automatically discover new fault modes and optimize diagnostic rules. The knowledge base supports online updates, allowing newly discovered fault modes and lessons learned to be incorporated into the diagnostic system in real time. The system also features a manual intervention interface, allowing operations and maintenance personnel to supplement expert experience and continuously enhance the system's intelligent diagnostic capabilities.

[0061] In an embodiment of the present invention, when the ground reference station module performs modeling compensation for regional atmospheric errors based on multi-base station joint solution, the process specifically includes the following: The system utilizes a distributed network of base station nodes to establish a spatial correlation model based on the Kriging interpolation algorithm, accurately describing the spatial distribution characteristics of ionospheric and tropospheric delays within a region. Each base station is equipped with a dual-frequency BeiDou receiver and meteorological sensors, collecting raw observation data and meteorological parameters at a 1Hz frequency. This data is then shared and jointly calculated in real time via an inter-station communication network.

[0062] The system constructs a state-space model that includes key parameters such as the total ionospheric electron count and tropospheric wet delay, and estimates parameter changes in real time using redundant observations from a network of reference stations. An adaptive noise covariance adjustment mechanism is incorporated into the filtering process, dynamically optimizing estimation weights based on satellite geometry and observation quality. To address the diurnal variability of the ionosphere, the system uses different sets of process noise parameters to ensure accurate delay estimates both during the day and at night.

[0063] A dedicated data compression algorithm is designed to encapsulate atmospheric correction parameters, orbit corrections, and integrity information into compact data frames. AES-256 encryption is used during transmission to ensure data security, and digital signatures are added to prevent data tampering. The system supports 4G / 5G multi-mode communication and automatically selects the optimal transmission path based on network conditions, ensuring the real-time and reliable broadcast of corrections.

[0064] A multi-reference station consistency check algorithm identifies anomalous station data, and robust estimation methods are used to mitigate the impact of outliers on the joint solution. The system calculates the observation residuals and stability indicators for each reference station in real time, automatically downgrading or removing stations that exceed thresholds. It also monitors satellite signal multipath and cycle slips, providing comprehensive data quality indicators to the drone, assisting in the selection and weighting of observations.

[0065] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0066] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A centimeter-level hovering control system for UAVs based on single Beidou differential positioning, characterized in that: The system includes a BeiDou RTK high-precision positioning module, a multi-sensor data fusion module, a flight control module, a motor drive and execution module, an exception handling and fault tolerance module, and a ground reference station module, wherein: The BeiDou RTK high-precision positioning module is used to perform real-time dynamic carrier phase differential processing on the UAV positioning signal based on the original BeiDou satellite observation data and the differential correction number provided by the ground reference station, thereby generating real-time position coordinates with centimeter-level accuracy; The multi-sensor data fusion module is used to use the tightly coupled filtering algorithm and the centimeter-level position data provided by the Beidou RTK module to perform multi-source data fusion on the IMU raw measurement values ​​and barometric altitude information, and ultimately output a high-reliability pose estimate with high-frequency updates; The flight control module is used to implement rolling time domain optimization control of the UAV motion state using the model predictive control algorithm and the posture data provided by the multi-sensor fusion module to generate the motor control quantity within the cycle; The motor drive and execution module is used to precisely drive the brushless motor through PWM modulation technology and control instructions issued by the flight control module to achieve millisecond-level response power output; The exception handling and fault tolerance module is used to diagnose and grade system anomalies using a multi-dimensional health monitoring algorithm and real-time operation data of each module, and output downgrade operation instructions or emergency protection signals; The ground reference station module is used to perform real-time differential calculations on satellite observation errors based on a high-precision Beidou reference receiver and known reference station coordinates, and continuously broadcast differential correction numbers with millimeter-level accuracy through a 4G communication link.

2. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: The BeiDou RTK high-precision positioning module, when performing filtering optimization on pseudo-range observation noise based on carrier phase smoothing technology, includes: Relying on the carrier phase continuity feature, the pseudo-range observation value is processed by sliding average to obtain a smooth observation. Measure sequence; Combined with the satellite elevation angle parameters, the filtering weights are dynamically allocated to form the optimal weighted result; The cycle slip detection algorithm is used to automatically remove abnormal data to ensure observation continuity.

3. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: The BeiDou RTK high-precision positioning execution module performs real-time correction of satellite orbit errors based on the base station differential correction number, including: The dual-frequency observation combination is used to separate the ionospheric delay error and generate the ionospheric correction. With the help of meteorological observation data, the tropospheric refraction is compensated by the model and the atmospheric correction parameters are output; The satellite orbit deviation is differentially eliminated using the precise coordinates of the reference station to generate a position correction result.

4. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: When the multi-sensor data fusion module performs alignment processing on heterogeneous sensor data based on a time synchronization mechanism, it includes: Use hardware trigger signals to synchronize multi-source sensor data and establish a unified time base; Relying on the interpolation compensation algorithm, the asynchronous sampling data is time-aligned to form a synchronous observation sequence.

5. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: The multi-sensor data fusion module, when performing optimal weighting of pose estimation errors based on a tightly coupled filtering algorithm, includes: Adaptive Kalman filtering is used to complete the covariance estimation of multi-source observation errors and construct the optimal weight matrix; Combined with the residual detection mechanism, abnormal observation data can be automatically eliminated to ensure the reliability of the fusion results.

6. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: The flight control module, when performing the prediction of the six-degree-of-freedom motion of the UAV based on dynamic modeling, includes: The rigid body dynamics model is used to predict the UAV's motion state and generate a state estimate. The aerodynamic parameter library is used to complete the modeling and compensation of wind disturbance effects and output anti-disturbance prediction results.

7. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: When the flight control module performs closed-loop adjustment on the posture tracking deviation based on the error feedback mechanism, it includes: Relying on proportional-integral-differential control, the posture deviation is gradually eliminated to generate a steady-state control variable; The feedforward compensation technology is used to dynamically correct the system lag effect and form a fast response instruction.

8. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: When the motor drive and execution module performs power conversion on the control instruction based on PWM modulation technology, it includes: Use space vector modulation to synthesize the waveform of three-phase voltage instructions and generate drive pulse signals; The dead-time compensation technology is used to minimize switching losses and ensure power conversion efficiency.

9. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1 is characterized in that: When the exception handling and fault tolerance module performs real-time monitoring of sensor data quality based on health assessment, it includes: Use the chi-square test method to quickly detect abnormal sensor data and trigger a fault alarm; Relying on the voting decision-making mechanism, the redundant system can achieve mode switching and maintain basic operating functions.

10. The centimeter-level hovering control system for unmanned aerial vehicles based on single Beidou differential positioning according to claim 1, characterized in that: When the ground reference station module performs modeling compensation for regional atmospheric errors based on multi-base station joint solution, it includes: Use network RTK technology to complete spatial modeling of regional atmospheric errors and construct error correction fields; The differential correction data can be broadcast in real time through the data broadcast protocol to provide positioning enhancement services.

Images