Multi-uav cooperative hoisting system and dynamic redundancy control method thereof

Abstract

The application discloses a multi-unmanned aerial vehicle cooperative hoisting system and a dynamic redundancy control method thereof, and belongs to the technical field of unmanned aerial vehicle cooperative operation. The system comprises eight unmanned aerial vehicles, which are divided into a first layer main working unit (four unmanned aerial vehicles) and a second layer hot backup unit (four unmanned aerial vehicles), and are connected to a load platform through a rigid structure with an intelligent bearing connecting device. The intelligent bearing connecting device can rotate freely under normal conditions to buffer internal stress, and can be controlled to be locked to provide a rigid reference when fault switching. A central controller monitors the health status of each machine in real time, and uses machine learning to predict faults; when a single machine of the main working unit fails, a standby machine is activated from the hot backup unit, and the thrust of each machine is reconstructed in real time through a dynamic control distribution algorithm to ensure that the resultant force passes through the center of mass. The application realizes whole-machine hot backup and dynamic reconstruction at the unmanned aerial vehicle level, and significantly improves the reliability and fault tolerance of the multi-machine cooperative hoisting system.

Description

Technical Field

[0001] This invention belongs to the field of UAV collaborative control and aerial transportation technology, and particularly relates to a multi-UAV collaborative lifting system and its control method that achieves heavy-load lifting through rigid connection of multiple UAVs and has dynamic hot backup and fault self-reconfiguration capabilities. Background Technology

[0002] With the rapid development of multi-rotor drone technology, the collaborative lifting of large or heavy-load materials exceeding the carrying capacity of a single drone has become a research hotspot. In existing technologies, collaborative lifting of multiple drones mainly employs the following two connection methods: (a) Flexible cable connection method. Multiple UAVs are connected to the same load via cables, and the lifting is achieved by coordinating the pulling force of each UAV. Although this method has a certain passive sway reduction effect, the flexibility of the cable leads to low load attitude control accuracy, and when a UAV fails, the system is difficult to reconfigure quickly, which can easily lead to load skewing or even falling.

[0003] (II) Rigid Connection Method. Multiple UAVs are connected into a single flight platform via a rigid structure. For example, Chinese patent application CN103699130A discloses a multi-UAV coordinated lifting device that connects multiple UAVs to a load via a boom; Chinese patent application CN121291770A discloses a multi-UAV collaborative heavy-duty lifting system that uses a modular design to connect a power module via a load-bearing bracket. In addition, there are existing technologies that rigidly connect two UAVs at a central position. However, existing rigid connection methods have the following shortcomings: 1. Lack of effective internal stress buffering mechanism: The rigid connection between UAVs means that any attitude disturbance of a single UAV will be directly transmitted to other UAVs through the rigid structure, generating harmful internal stress and increasing the risk of structural fatigue and loss of control. Although there are existing technologies that interconnect UAVs through semi-rigid connections (such as ball joints), they only provide passive flexibility and lack the ability to actively lock rigidly in fault conditions.

[0004] 2. Insufficient Redundancy Backup Capability: Current solutions involve the entire UAV in operation, lacking an effective whole-machine-level hot backup mechanism. Existing redundancy technologies are mostly limited to the flight control computer level. For example, Chinese patent CN111352338A discloses a dual-redundant flight control computer employing similar redundancy and hot backup for the main and auxiliary units; other technologies involve hot backup of servo controllers. However, these are all component-level redundancy; no technical solution has yet been found that uses the entire UAV as a replaceable power unit for hot backup.

[0005] 3. Lagging fault switching response: Existing redundancy switching technologies mostly rely on passive detection and switching after a fault occurs, and there are obvious thrust fluctuations and attitude disturbances during the switching process.

[0006] 4. Limited control and allocation strategy: Existing solutions lack flexible dynamic control and allocation capabilities after a failure occurs, making it difficult to optimize pull allocation in real time based on the spatial location of the remaining available UAVs.

[0007] Therefore, there is an urgent need in this field for a multi-UAV collaborative hoisting system that can leverage the strength advantages of rigid connection structures, effectively buffer internal stress, and possess whole-machine-level hot backup and dynamic reconfiguration capabilities. Summary of the Invention

[0008] (a) Purpose of the invention This invention aims to address the shortcomings of the prior art by providing a multi-UAV collaborative hoisting system and its dynamic redundancy control method. Through a "4+4" hierarchical hot backup architecture, a lockable intelligent bearing connection device, and an event-triggered dynamic control allocation algorithm, the multi-UAV collaborative hoisting system can achieve rapid reconfiguration and stable flight under single-UAV failure conditions.

[0009] (II) Technical Solution 1. System Overall Architecture This invention provides a multi-UAV collaborative hoisting system, comprising: The first layer of the drone swarm consists of 4 drones, which serve as the main working units. They are located on the first layer above or below the load platform and are fixedly connected to the load platform through a rigid connection structure. They bear the main lifting thrust. The second-layer drone swarm consists of four drones, which serve as a hot backup unit. It is located on the second layer, which is parallel to the first layer, and is connected to the payload platform through a rigid connection structure. Loading platform: Used to support or suspend heavy objects to be hoisted; Rigid connection structure: includes several connecting rods and an intelligent bearing connection device set between the connecting rods and the drone; Central Controller: Communicates with all drones to monitor their status in real time, perform fault detection and prediction, and dynamically allocate control. Status awareness module: includes positioning sensors, tension sensors, vibration sensors, current sensors and temperature sensors installed on each UAV.

[0010] The first-layer drone swarm and the second-layer drone swarm are arranged vertically in space, with four drones in each layer positioned at the four corners of the payload platform, forming a symmetrical quadrilateral layout.

[0011] 2. Intelligent bearing connection device This invention provides a rotatable and controllable locking intelligent bearing connection device, comprising: The bearing body includes an inner ring, an outer ring, and rolling elements. The inner ring is fixedly connected to the drone connection end, and the outer ring is fixedly connected to the connecting rod, enabling relative rotation between the drone and the connecting rod. Locking mechanism: Located inside or outside the bearing body, including an electromagnetic actuator and a locking element. When a locking command is received, the inner ring and the outer ring are locked relative to each other, so that the bearing switches from a rotatable state to a rigid fixed state. Unlocking mechanism: Releases the lock upon receiving an unlocking command, allowing the bearing to return to a rotatable state; Angle sensor: used to detect the relative rotation angle and angular velocity between the inner and outer rings of the bearing in real time; Locking controller: Communicates with the central controller and controls the locking and unlocking of the locking mechanism according to the instructions of the central controller.

[0012] The locking mechanism may employ electromagnetic locking, mechanical locking, or a combination of both, with a locking response time of no more than 50 milliseconds.

[0013] While existing technologies include spherical UAV solutions that place the locking mechanism inside the bearing, and rotational self-locking connection devices on UAV landing gear, their application scenarios and functional purposes are different from those of this invention. The former is used for the roll control of spherical UAVs, and the latter is used for landing gear retraction and fixation. Neither of them involves the switching between "normal flexible buffer - fault rigid locking" modes at the rigid connection nodes of multiple aircraft.

[0014] 3. Dynamic Redundancy Control Method This invention provides a dynamic redundancy control method for a multi-UAV collaborative hoisting system, comprising the following steps: S1: System Initialization and Health Assessment After the system is powered on, the central controller performs a self-check on all eight drones, obtains the health status information of each drone, and establishes a global health status mapping table. Based on the health status, the drones are divided into main working units (4 units) and hot backup units (4 units).

[0015] S2: Normal operating mode The four drones in the main working unit simultaneously output thrust, hoisting the load to the target location via a rigid connection structure. The four drones in the hot backup unit synchronously receive control commands from the central controller, maintaining the same attitude and rotation speed commands as the main working unit, but output zero or minimal thrust (only enough to maintain their own hovering), remaining in a "hot standby" state. The intelligent bearing connection device is in an unlocked (rotatable) state, allowing each drone to rotate freely relative to the connection structure, buffering attitude disturbances and internal stresses during flight.

[0016] S3: Predictive Health Monitoring The central controller collects characteristic parameters of each UAV in real time, such as vibration spectrum, current waveform, temperature change rate, and positioning deviation, through the state awareness module. It uses machine learning algorithms (such as support vector machines and long short-term memory networks) to perform online assessment and fault prediction of the UAV's health status. When a fault is predicted to occur in a UAV, a pre-switching procedure is initiated before the fault actually occurs.

[0017] S4: Fault Detection and Event Triggering When a malfunction or health level below a preset threshold is detected in a drone within the main working unit, an event response is triggered: S4.1: The central controller immediately sends a locking command to the intelligent bearing connection device corresponding to the faulty UAV, instantly locking the connection point into a rigid state and providing an instantaneous rigidity reference for the system; S4.2: The central controller selects a replacement drone from the hot backup unit based on the spatial location of the faulty drone; S4.3: If there is a hot backup drone in a spatially symmetrical position to the faulty drone, the backup drone in that symmetrical position shall be activated first (symmetry reconfiguration). S4.4: If the backup drone at the symmetrical location is unavailable, activate the hot backup drone at any location (asymmetric reconfiguration).

[0018] S5: Dynamic Control Allocation and Reconfiguration The central controller executes a dynamic control allocation algorithm to recalculate the expected pull for all participating drones (including the remaining 3 drones in the original main working unit and the 1 active hot backup drone): S5.1: Establish a spatial coordinate system with the payload platform's center of mass as the origin, and obtain the position vector ri and the current maximum available thrust T of each UAV. max ; S5.2: Establish a control allocation optimization model, with the objective function being to minimize the weighted sum of total thrust deviation and attitude moment deviation; S5.3: Constraints include: the resultant force is equal to the desired total thrust (≥ the load gravity), the resultant moment is zero (or equal to the desired attitude moment), and the thrust of each engine does not exceed the maximum available thrust; S5.4: Use the weighted pseudo-inverse method or quadratic programming algorithm to solve for the optimal thrust allocation scheme; S5.5: Send the calculated expected thrust commands for each aircraft to the corresponding UAV, and complete the thrust reconfiguration at the moment of fault switching.

[0019] S6: Refactoring Confirmation and Continuous Monitoring The central controller monitors the system attitude and load swing amplitude in real time after reconfiguration. If the system stabilizes within a preset time (e.g., 0.5 seconds), the reconfiguration is confirmed as successful, and the hoisting task continues; if the reconfiguration fails or a new fault occurs, the emergency landing procedure is initiated.

[0020] 4. Optimal technical solution Preferably, the four drones of the hot backup unit rotate synchronously with the main working unit under normal conditions, ensuring that they can quickly intervene with minimal attitude difference when a switch is required.

[0021] Preferably, the dynamic control allocation algorithm adopts a model predictive control framework, which pre-calculates the optimal reconfiguration scheme under multiple fault scenarios before the fault occurs, thereby further shortening the fault response time.

[0022] (III) Beneficial Effects Compared with the prior art, the present invention has the following advantages: 1. Whole-machine-level hot backup significantly improves system reliability: This invention is the first to use four complete UAVs as hot backup units, achieving power redundancy at the UAV level and overcoming the limitations of existing technologies that can only achieve redundancy at the flight control or sensor level (such as CN111352338A). When the main working unit fails, the hot backup UAV can immediately take over, and the total thrust of the system will not fluctuate drastically.

[0023] 2. Lockable intelligent bearing for rigid-flexible switching: The intelligent bearing connection device provided by this invention is rotatable under normal conditions, effectively buffering attitude disturbances and internal stresses during flight; it locks immediately upon fault switching, providing an instantaneous rigid reference point for control reconfiguration, thus balancing the dual requirements of "flexible buffering" and "rigid control." Unlike existing spherical UAV bearing locking technology and landing gear locking devices, this invention is the first to apply it to dynamic mode switching of multi-aircraft rigid connections.

[0024] 3. Event-triggered dynamic control allocation, balancing computational efficiency and control accuracy: This invention adopts an event-triggered strategy of "simplified control under normal conditions and complex optimization under fault conditions". Under normal conditions, only simple PID control is required. Only when a fault occurs is a complex nonlinear optimization algorithm triggered, which greatly reduces the consumption of computational resources while ensuring control accuracy.

[0025] 4. Predictive health management for proactive fault tolerance: This invention introduces a fault prediction mechanism based on machine learning, which can start a pre-switch procedure before a fault occurs, upgrading the traditional "passive response" fault tolerance to "proactive prevention" fault tolerance, significantly improving the system's security and task success rate.

[0026] 5. Symmetrical and asymmetric dual reconfiguration strategy: This invention supports both rapid replacement of standby machines in symmetrical locations and flexible allocation of standby machines in arbitrary locations, greatly improving the system's adaptability to complex fault scenarios. Attached Figure Description

[0027] Figure 1 The overall structural diagram (front sectional view) of the multi-UAV collaborative hoisting system of this invention shows the spatial arrangement of the first-layer UAV swarm (4 units), the second-layer UAV swarm (4 units), the load platform, and the rigid connection structure.

[0028] Figure 2 The cross-sectional structural diagram of the intelligent bearing connection device of the present invention shows the internal structure and connection relationship of the bearing body, locking mechanism, angle sensor and locking controller.

[0029] Figure 3 : A comparative schematic diagram of the intelligent bearing connection device of the present invention in the unlocked state (rotatable) and the locked state (rigidly fixed).

[0030] Figure 4 The flowchart of the dynamic redundancy control method of this invention shows the complete control logic from system initialization, normal operation, predictive monitoring, fault detection and dynamic reconfiguration.

[0031] Figure 5 : A schematic diagram of the symmetry reconstruction strategy of the present invention, showing the spatial correspondence between the faulty UAV and the standby UAV in a symmetrical position.

[0032] Figure 6 : A schematic diagram of the asymmetric reconfiguration strategy of the present invention, showing the control allocation vector diagram after the standby machine is activated at any position.

[0033] Figure 7 : A schematic diagram comparing the thrust distribution of the system before and after fault switching. Detailed Implementation

[0034] Example 1 This embodiment provides a multi-drone collaborative lifting system, including eight quadcopter drones (each with a maximum lifting capacity of 50 kg), a load platform, and a rigid connection structure connecting the eight drones to the load platform. The eight drones are arranged in two layers, with four drones in each layer, and are fixed to the connection points at the four corners of the load platform. Each drone is connected to the connecting rod via the intelligent bearing connection device described in this invention.

[0035] The intelligent bearing connection device uses an angular contact ball bearing as the bearing body, and an electromagnetic locking mechanism is located between the outer and inner rings of the bearing. Under normal conditions, the electromagnetic locking mechanism is de-energized, and the bearing can rotate freely; when a locking command is received, the electromagnetic locking mechanism is energized, and the locking element extends to lock the inner and outer rings together.

[0036] The central controller uses an embedded industrial computer to communicate with each UAV via the 5.8GHz frequency band, with a communication delay of no more than 5 milliseconds. Each UAV is equipped with an RTK-GNSS positioning module (positioning accuracy ±2cm), a tension sensor (range 0-100kg, accuracy ±0.5%), a triaxial vibration sensor (frequency response range 0-500Hz), and a current sensor.

[0037] Example 2 This embodiment provides a specific implementation method for dynamic redundancy control.

[0038] During normal operation: 4 main working drones hover and carry loads at 80% of their rated thrust, while 4 hot backup drones maintain their hovering attitude at 5% of their rated thrust. All intelligent bearings are unlocked. The central controller collects vibration, current, and temperature data from all drones at a frequency of 100Hz and inputs this data into a pre-trained Long Short-Term Memory (LSTM) network model for health status prediction.

[0039] Fault detection phase: When abnormal characteristics appear in the vibration spectrum of a main working drone and the health prediction value continues to drop below the threshold for three consecutive sampling periods, the central controller determines that the drone is about to fail and triggers the pre-switching procedure.

[0040] Switching Phase: Within 50 milliseconds, the central controller sends a locking command to the intelligent bearing corresponding to the faulty drone. Simultaneously, it selects a backup drone from the hot backup unit that is spatially symmetrical to the faulty drone and sends an activation command to it (increasing the target thrust from 5% to 80%). The locking mechanism completes the locking action within 30 milliseconds.

[0041] Reconfiguration Phase: The central controller executes a weighted pseudo-inverse control allocation algorithm to calculate the optimal thrust allocation scheme for the four working drones (three original main drones + one activated backup drone) within 20 milliseconds and issues the scheme for execution. The system completes the entire process from fault detection to attitude recovery within 150 milliseconds, with the load swing amplitude controlled within ±3°.

Claims

1. A multi-UAV collaborative hoisting system and its dynamic redundancy control method, characterized in that, include: The first layer of the drone swarm consists of four drones and is located on the first layer. It is connected to the payload platform through a rigid connection structure. The second-layer drone swarm consists of four drones and is located on the second layer, which is parallel to the first layer. It is connected to the load platform through a rigid connection structure. A load platform is used to support or suspend heavy objects to be lifted. The rigid connection structure includes several connecting rods and an intelligent bearing connection device disposed between the connecting rods and the drone; The central controller communicates with all drones and is used to monitor the status of each drone in real time and perform dynamic control allocation. The status awareness module includes positioning sensors, tension sensors, vibration sensors, current sensors, and temperature sensors installed on each UAV; Under normal conditions, the central controller controls the four drones in the first-layer drone swarm to output main thrust, and the four drones in the second-layer drone swarm to output zero thrust or minimal thrust as hot backup units. When a failure is detected in any drone in the first-layer drone swarm, the central controller activates at least one drone from the second-layer drone swarm to take over the failed drone and recalculates the thrust distribution of all working drones so that the combined force passes through the center of mass of the load platform.

2. The multi-UAV collaborative hoisting system according to claim 1, characterized in that, The intelligent bearing connection device includes: The bearing body includes an inner ring, an outer ring, and rolling elements. The inner ring is fixedly connected to the connection end of the UAV, and the outer ring is fixedly connected to the connecting rod. A locking mechanism, located inside or outside the bearing body, includes an actuator and a locking element, which locks the inner ring relative to the outer ring when a locking command is received; The unlocking mechanism releases the lock upon receiving an unlocking command; An angle sensor is used to detect the relative rotation angle and angular velocity between the inner and outer rings of the bearing in real time. The locking controller communicates with the central controller and controls the locking and unlocking of the locking mechanism according to the instructions of the central controller.

3. The multi-UAV collaborative hoisting system according to claim 2, characterized in that, The locking mechanism is an electromagnetic locking mechanism or a mechanical locking mechanism, and the locking response time is no more than 50 milliseconds.

4. The multi-UAV collaborative hoisting system according to claim 1, characterized in that, The first-layer drone swarm and the second-layer drone swarm are arranged vertically in space, with four drones in each layer positioned at the four corners of the payload platform, forming a symmetrical quadrilateral layout.

5. A dynamic redundancy control method for a multi-UAV collaborative hoisting system, applied to the system described in any one of claims 1-4, characterized in that, Includes the following steps: S1: System initialization, the central controller performs a self-check on all drones, and divides the 8 drones into 4 main working units and 4 hot backup units according to their health status. S2: Normal operation, the four drones of the main working unit output the main thrust, the four drones of the hot backup unit receive control commands synchronously but output zero thrust or very small thrust, and the intelligent bearing connection device is in the unlocked state. S3: Predictive health monitoring. The central controller collects vibration, current, temperature and positioning data of each UAV in real time, and uses machine learning algorithms to conduct online assessment of the health status of the UAV and predict faults. S4: When a drone in the main working unit is detected to have malfunctioned or its health level is lower than a preset threshold, the central controller sends a locking command to the corresponding intelligent bearing connection device and selects a drone from the hot backup unit as a substitute. S5: The central controller executes a dynamic control allocation algorithm to recalculate the expected pull of all participating UAVs, ensuring that the resultant force passes through the center of mass of the load platform and the resultant torque is zero. S6: The central controller sends the calculated expected thrust commands of each drone to the corresponding drone to complete the thrust reconfiguration.

6. The method according to claim 5, characterized in that, Step S4, which involves selecting a drone from the hot backup unit as a substitute, specifically includes: Prioritize selecting a hot backup drone that is spatially symmetrical to the faulty drone on the payload platform; If the hot backup drone at the symmetrical location is unavailable, select the hot backup drone at any location.

7. The method according to claim 5, characterized in that, The dynamic control allocation algorithm described in step S5 includes: Establish a spatial coordinate system with the payload platform's center of mass as the origin, and obtain the position vector ri and the current maximum available thrust Ti of each UAV. max ; A control allocation optimization model is established, with the objective function being to minimize the weighted sum of total thrust deviation and attitude moment deviation; The constraints include the resultant force being equal to the desired total thrust, the resultant moment being zero or equal to the desired attitude moment, and the thrust of each engine not exceeding the maximum available thrust. The optimal thrust allocation scheme is solved by using the weighted pseudo-inverse method or quadratic programming algorithm.

8. The method according to claim 5, characterized in that, The machine learning algorithm includes support vector machines or long short-term memory networks, used to predict the health status of the UAV based on its vibration spectrum, current waveform, temperature change rate, and positioning deviation.

9. The method according to claim 5, characterized in that, The intelligent bearing connection device is normally in an unlocked state, allowing free rotation. During fault switching, it is locked into a rigid state, and after reconstruction, it is restored to the unlocked state.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 5-9.

Images