Large unmanned aerial vehicle intelligent flight control integrated system

Abstract

The invention provides an intelligent flight control integrated system for a large unmanned aerial vehicle, and the system comprises a multi-mode environment sensing module, a self-adaptive flight control module, a dynamic power management module, a distributed communication architecture module, an intelligent task decision module, and a fault self-repairing system. The multi-mode environment sensing module is connected with the adaptive flight control module through a bidirectional real-time data bus, and transmits environment sensing data to the intelligent task decision module through an optical fiber communication redundant link; the dynamic power management module is connected with each function module through an intelligent power distribution matrix, and the intelligent power distribution matrix comprises a 16-channel programmable MOSFET array. Through cross-module collaborative design (perception-control-energy-communication) and physical constraint embedding (moment balance and optical fiber mechanical protection), the system has comprehensive advantages in the aspects of environmental adaptability, energy efficiency management, communication reliability, extreme fault tolerance and the like, and the performance boundary of a traditional unmanned aerial vehicle system is remarkably exceeded.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent flight control of unmanned aerial vehicles, and specifically relates to an intelligent flight control integration system for large unmanned aerial vehicles. Background Art

[0002] An unmanned aerial vehicle is short for an unmanned aircraft (Unmanned Aerial Vehicle), which is an unmanned aircraft using radio remote control equipment and self - contained program control devices, including unmanned helicopters, fixed - wing aircraft, multi - rotor aircraft, unmanned airships, and unmanned parafoil aircraft. Broadly speaking, it also includes near - space aircraft (in the airspace of 20 - 100 kilometers), such as stratospheric airships, high - altitude balloons, and solar unmanned aerial vehicles. From a certain perspective, an unmanned aerial vehicle can complete complex aerial flight tasks and various load tasks under unmanned conditions and can be regarded as an "aerial robot".

[0003] The current flight control systems of large unmanned aerial vehicles still have the following defects:

[0004] 1. Insufficient environmental perception: Traditional unmanned aerial vehicles rely on a single sensor (such as a camera or GPS), and are vulnerable to noise in complex environments (such as strong light, rain, fog, and electromagnetic interference), resulting in low target recognition accuracy and a high failure rate of obstacle avoidance.

[0005] 2. Rigid control strategy: Control algorithms with fixed parameters are difficult to adapt to dynamic scenarios (such as gust disturbances and task changes), with significant response delays and poor flight stability.

[0006] 3. Inefficient energy management: The single battery pack design is prone to thermal runaway, and lacks an aging prediction mechanism, resulting in short endurance and low lifespan.

[0007] 4. Poor communication reliability: Traditional wireless communication is vulnerable to interference, and the risk of link interruption is high in extreme environments, affecting task continuity.

[0008] 5. Weak fault tolerance: Redundant designs are mostly limited to single - component backups (such as dual motors) and cannot cope with extreme faults such as the failure of multiple propellers. Summary of the Invention

[0009] The present invention aims to solve the problems raised in the background art and provides an intelligent flight control integration system for large unmanned aerial vehicles.

[0010] The specific technical solution is as follows: An intelligent flight control integration system for large unmanned aerial vehicles includes: a multi - modal environmental perception module, an adaptive flight control module, a dynamic power management module, a distributed communication architecture module, an intelligent task decision module, and a fault self - repair system;

[0011] The multi-modal environment perception module is connected to the adaptive flight control module through a bi-directional real-time data bus, and transmits environment perception data to the intelligent task decision module via a fiber optic communication redundant link;

[0012] The dynamic power management module is connected to each functional module through an intelligent power distribution matrix, and the intelligent power distribution matrix includes a 16-channel programmable MOSFET array, supporting μs-level power dynamic distribution;

[0013] The distributed communication architecture module connects the master control node and at least three redundant sub-nodes through a star topology network, and a time-division multiplexing + frequency-division multiplexing hybrid communication protocol is adopted between the nodes;

[0014] The fault self-repair system is interconnected with all actuators through a dual-redundancy CAN bus, and is embedded in the load monitoring interface of the dynamic power management module to achieve current-temperature joint diagnosis.

[0015] In the above large UAV intelligent flight control integration system, the multi-modal environment perception module includes:

[0016] A lidar unit, a millimeter-wave radar unit, a visible light camera unit, and an infrared sensing unit, and the data of each unit are synchronously fused at the hardware level through an FPGA chip;

[0017] Equipped with a real-time error compensation system based on an improved Kalman filter, and its dynamic weight distribution function is:

[0018]

[0019] Where: λ i 、λ j are both attenuation factors (value range 0.5 - 2);

[0020] Δz i (t) is the measurement residual of the i-th sensor at time t;

[0021] Δz j (t) is the measurement residual of the j-th sensor at time t;

[0022] τ is the sensor response delay compensation amount (0 ≤ τ ≤ 50ms);

[0023] is the noise variance of sensor i after delay compensation, dynamically predicting the noise level at the future τ moment.

[0024] In the above large UAV intelligent flight control integration system, the adaptive flight control module includes an attitude controller based on deep reinforcement learning, and its reward function is:

[0025]

[0026] Among them: k1, k2, and k3 are dynamically adjusted according to the flight mode: the weight in the cruise stage is (0.7, 0.2, 0.1), and in the obstacle avoidance stage is (0.4, 0.1, 0.5). Suppress the jitter of the control command, 0.01 ≤ γ ≤ 0.1; V is the current real-time flight speed of the UAV (unit: m / s); Vmax is the maximum designed flight speed of the UAV (unit: m / s); P 电机 ( t): The real-time power of the motor is detected through the current-voltage closed loop.

[0027] For the above-mentioned intelligent flight control integration system of large UAVs, among them, the dynamic power management module includes:

[0028] A multi-battery pack parallel topology structure, and a bidirectional DC-DC converter is used to achieve dynamic energy allocation;

[0029] A battery health status evaluation unit, which calculates the aging factor based on the electrochemical impedance spectrum:

[0030]

[0031] Among them: SOH i is the health status of the i-th battery pack; C 实际 is the current actual available capacity of the battery pack (unit: Ah or mAh); C 标称 is the rated capacity of the battery at the time of factory (unit: Ah or mAh); μ is the aging attenuation coefficient; N is the number of complete charge and discharge cycles experienced by the battery pack (unit: times);

[0032] The objective function of the temperature-load joint optimization strategy of the dynamic power management module is extended to:

[0033]

[0034] Among them: α, β, and η are weight coefficients; T max is the real-time maximum temperature of the battery pack (unit: °C); T opt is the dynamic optimal operating temperature (unit: °C); P i is the actual output power of the i-th battery pack (unit: W); P req is the power currently required by the system (unit: W); P rated is the rated power of the battery pack (unit: W).

[0035] For the above-mentioned intelligent flight control integration system of large UAVs, among them, the distributed communication architecture module adopts:

[0036] An adaptive frequency hopping communication protocol, which dynamically switches the frequency band according to the channel quality;

[0037] Quantum encryption transmission unit, supporting BB84 protocol key distribution;

[0038] Redundant optical fiber communication link, satisfying the mechanical constraint equation:

[0039]

[0040] where dL / dt is the optical fiber release rate, which is 0.5·V in the emergency obstacle avoidance mode max ; θ bend ≤15°, the optical fiber bending angle, and when it exceeds 15°, the automatic retraction protection mechanism is triggered; Vmax is the maximum designed flight speed of the unmanned aerial vehicle (unit: m / s).

[0041] The above-mentioned intelligent flight control integrated system for large unmanned aerial vehicles, wherein

[0042] The intelligent task decision-making module includes:

[0043] Multi-objective optimization decision-making engine, using the NSGA-II algorithm to balance task efficiency and energy consumption;

[0044] Task rehearsal subsystem based on digital twin, simulating the flight environment through a high-precision dynamics model;

[0045] Human-machine collaborative decision-making interface, supporting natural language instruction parsing and multi-modal intention recognition.

[0046] The above-mentioned intelligent flight control integrated system for large unmanned aerial vehicles, wherein the fault self-repair system includes:

[0047] Dual-redundant actuator, adopting a heterogeneous drive design with electromagnetic servo and hydraulic servo in parallel;

[0048] Health monitoring network for key components, detecting abnormal frequency components based on vibration spectrum analysis;

[0049] Dynamic control system reconstruction algorithm, maintaining stability through torque redistribution when some propellers fail, satisfying:

[0050] where: F i is the lift force of the i-th propeller (unit: N); R i is the arm length of the i-th propeller to the center of gravity of the unmanned aerial vehicle (unit: m); I is the moment of inertia of the airframe (unit: kg·m 2 ); θ is the angular acceleration of the airframe (unit: rad / s 2 ).

[0051] The above-mentioned intelligent flight control integrated system for large unmanned aerial vehicles, wherein the system is configured with an anti-interference navigation and positioning module, including:

[0052] Multi-mode satellite positioning receiver;

[0053] Visual SLAM positioning unit, based on ORB feature matching algorithm;

[0054] Abnormal data cross-validation unit, using Mahalanobis distance detection:

[0055]

[0056] When D M > 3σ, trigger data correction;

[0057] Where: D M is the Mahalanobis distance; x is the current sensor data vector; μ is the mean vector of historical data; Σ is the covariance matrix; T is the transpose operator.

[0058] The above large-scale UAV intelligent flight control integration system, wherein the system adopts a modular and scalable architecture, including:

[0059] A hardware interface that supports hot plugging and complies with the ARINC810 standard;

[0060] An open software protocol stack that provides API interfaces for third-party algorithm integration;

[0061] In-air software upgrade capability, realizing firmware incremental update through differential compression technology, with a compression rate ≥ 70%.

[0062] The above large-scale UAV intelligent flight control integration system, wherein the application method in an extreme failure scenario:

[0063] When three propellers are detected to fail, trigger the passive disaster tolerance control algorithm, generate equivalent lift through the periodic variable speed of the remaining propellers, and satisfy the angular momentum conservation equation:

[0064]

[0065] Achieve a safe return;

[0066] Where, J is the body moment of inertia; w is the instantaneous angular velocity of the body (unit: rad / s); ΔL i is the lift increment of the i-th remaining propeller (unit: N); r i is the arm length of the i-th remaining propeller to the center of gravity of the UAV (unit: m); k is the number of remaining available propellers.

[0067] The present invention has the following beneficial effects:

[0068] 1. Enhanced environmental perception and fusion ability

[0069] Multi-modal sensors (lidar, millimeter-wave radar, visible light camera, infrared sensor) significantly improve the target recognition accuracy and anti-interference ability in complex environments through hardware-level synchronous fusion and dynamic weight allocation algorithms; the real-time error compensation system suppresses burst noise to ensure the long-term stability of the sensed data.

[0070] 2. Intelligent Flight Control Optimization

[0071] The deep reinforcement learning model combines dynamic reward functions (tracking accuracy, energy consumption, obstacle avoidance success rate) to achieve adaptive decision-making in complex scenarios; the aerodynamic parameter real-time self-compensation optimizes the flight attitude, enhancing the stability and safety of high-maneuver tasks.

[0072] 3. Efficient Energy Management

[0073] The multi-battery pack dynamic deployment strategy combines temperature-load joint optimization to balance energy consumption and safety, extend battery life and reduce the risk of thermal runaway; the intelligent power distribution matrix supports μs-level power distribution to adapt to sudden load demands.

[0074] 4. Reliable Communication and Navigation

[0075] Quantum encryption and fiber optic redundant link design enhance the anti-interference ability of communication and ensure the continuity of high-bandwidth data transmission; multi-mode positioning and cross-verification of abnormal data enhance the navigation accuracy and reliability in complex environments.

[0076] 5. Multi-level Fault Tolerance

[0077] Heterogeneous redundant actuators and dynamic torque distribution algorithms quickly restore flight stability and support safe return in scenarios where multiple propellers fail; the health monitoring network diagnoses key component abnormalities in real time and gives early warnings of potential faults.

[0078] 6. Modular Expansion and Easy Maintenance

[0079] Hot-swappable hardware interfaces and open software protocol stacks support rapid function expansion and third-party integration; the in-air software upgrade ability reduces maintenance costs and improves the system iteration efficiency. Brief Description of the Drawings

[0080] Figure 1 It is a schematic diagram of the architecture of the intelligent flight control integration system for large unmanned aerial vehicles of the present invention;

[0081] Figure 2 It is a graph showing the change of the multi-modal sensor fusion accuracy over time of the present invention;

[0082] Figure 3 It is a graph showing the change of the control response delay with the task complexity of the present invention;

[0083] Figure 4It is a graph showing the change of the battery pack life of the present invention with the charge and discharge cycles;

[0084] Figure 5 It is a graph showing the change of the communication link reliability of the present invention with the load;

[0085] Figure 6 It is a graph showing the change of the positioning error of the present invention with time;

[0086] Figure 7 It is a graph showing the change of the fault recovery time of the present invention with the fault severity. Detailed implementation manners

[0087] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings and through specific implementation manners.

[0088] Among them, the accompanying drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and should not be construed as a limitation to this patent; in order to better illustrate the embodiments of the present invention, some components in the accompanying drawings will be omitted, enlarged or reduced, which do not represent the dimensions of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the accompanying drawings may be omitted.

[0089] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if terms such as "upper", "lower", "left", "right", "inner", "outer", etc. are used to indicate the orientation or position relationship, it is based on the orientation or position relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, so the terms describing the position relationship in the accompanying drawings are only for illustrative purposes and should not be construed as a limitation to this patent. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0090] In the description of the present invention, unless otherwise clearly defined and limited, if terms such as "connection" are used to indicate the connection relationship between components, this term should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0091] As Figures 1-7 shown, among which, Figure 1 It is a schematic diagram of the architecture of this system, showing the architecture of this system; Figure 2This is a graph showing the variation of the multi-modal sensor fusion accuracy of the system over time, demonstrating that the fusion accuracy of the multi-modal sensors gradually improves over time, indicating that the system's ability to perceive the environment is constantly enhancing; Figure 3 This is a graph showing the variation of the control response delay of the system with the task complexity, demonstrating that as the task complexity increases, the control response delay gradually decreases, indicating that the system can make decisions and responses more quickly; Figure 4 This is a graph showing the variation of the battery pack life of the system with the charge-discharge cycles, demonstrating that as the charge-discharge cycles increase, the battery pack life gradually decreases, but the rate of decrease gradually slows down, indicating that the system has a long service life and good battery management performance; Figure 5 This is a graph showing the variation of the communication link reliability of the system with the load, demonstrating that as the load increases, the communication link reliability gradually decreases, but always remains at a relatively high level, indicating that the system can maintain stable communication under high load; Figure 6 This is a graph showing the variation of the positioning error of the system over time, demonstrating that as time goes by, the positioning error gradually decreases, indicating that the navigation and positioning accuracy of the system is constantly improving; Figure 7 This is a graph showing the variation of the fault recovery time of the system with the fault severity, demonstrating that as the fault severity increases, the fault recovery time gradually lengthens, but always remains within an acceptable range, indicating that the system has strong fault tolerance and recovery capabilities. The large unmanned aerial vehicle intelligent flight control integration system provided in this embodiment is used for large unmanned aerial vehicle intelligent flight control. This system includes: a multi-modal environment perception module, an adaptive flight control module, a dynamic power management module, a distributed communication architecture module, an intelligent task decision module, and a fault self-repair system, where:

[0092] The multi-modal environment perception module is connected to the adaptive flight control module through a bidirectional real-time data bus, and transmits environment perception data to the intelligent task decision module via a fiber optic communication redundant link;

[0093] The dynamic power management module is connected to each functional module through an intelligent power distribution matrix, and the intelligent power distribution matrix includes a 16-channel programmable MOSFET array, supporting μs-level power dynamic distribution;

[0094] The distributed communication architecture module connects the master control node and at least three redundant sub-nodes through a star topology network, and a time-division multiplexing + frequency-division multiplexing hybrid communication protocol is adopted between the nodes;

[0095] The fault self-repair system is interconnected with all actuators through a dual-redundancy CAN bus, and is embedded in the load monitoring interface of the dynamic power management module to achieve current-temperature joint diagnosis.

[0096] With the above solution, through the collaborative connection of multiple modules (such as environmental perception, flight control, power management, communication architecture, etc.), the overall response speed and functional collaboration efficiency of the system are significantly improved; the two-way real-time data bus and fiber optic redundant link ensure the efficient transmission of environmental perception data, effectively reducing the risk of delay and packet loss; the μs-level power dynamic allocation based on the intelligent power distribution matrix can effectively improve energy utilization rate and support continuous power supply under complex tasks.

[0097] Among them, the multi-modal environmental perception module includes: lidar unit, millimeter-wave radar unit, visible light camera unit and infrared sensing unit, and the data of each unit is synchronously fused at the hardware level through the FPGA chip;

[0098] It also includes a real-time error compensation system equipped with an improved Kalman filter, and its dynamic weight distribution function is:

[0099]

[0100] Where: λ i 、λ j are both attenuation factors (the value range is 0.5 - 2), which suppress the burst noise of the sensor. Technical meaning: control the suppression intensity of the burst noise of the sensor, and the larger the value, the stronger the suppression of abnormal data;

[0101] Δz i (t) is the measurement residual of the i-th sensor at time t, that is, the deviation between the actual measurement value and the predicted value of the sensor i. (Δz i( t) = z i( t) - H·x(t|t - 1)), z i (t) is the original measurement value of the i-th sensor at time t; H is the observation matrix, which is used to map the system state to the sensor measurement space; x(t|t - 1) is the predicted value of the current moment based on the state estimation of the previous moment;

[0102] Δz j (t) is the measurement residual of the j-th sensor at time t, that is, the deviation between the actual measurement value and the predicted value of the sensor j. (Δz j( t) = z j (t) - H·x(t|t - 1)), zj(t) is the original measurement value of the j-th sensor at time t; H is the observation matrix, which is used to map the system state to the sensor measurement space; x(t|t - 1) is the predicted value of the current moment based on the state estimation of the previous moment;

[0103] τ is the sensor response delay compensation amount (0 ≤ τ ≤ 50ms), and the adaptation rule is: lidar τ = 5ms (fast response), visible light camera unit τ = 20ms (slow response);

[0104] is the noise variance of sensor i after time delay compensation, and dynamically predicts the noise level at the future τ moment.

[0105] Example: During the flight of a drone, lidar and camera simultaneously detect obstacles. Since the lidar has a fast response (τ = 5ms), it updates the noise variance first. The camera, due to the delay (τ = 20ms), needs to predict the future noise. Through dynamic weight allocation, the weight of lidar data is higher, and burst noise (such as strong light interfering with the camera) is suppressed.

[0106] Technical effect: The accuracy of sensor fusion is significantly improved, and the ability to suppress burst noise is significantly enhanced.

[0107] Working principle process:

[0108] 1. The data of each sensor is input into the FPGA chip for hardware-level synchronization;

[0109] 2. Calculate the measurement residuals Δz i (t) and Δz j (t);

[0110] 3. Input λ i 、λ j and dynamically allocate weights;

[0111] 4. Output the environmental perception result after weighted fusion.

[0112] Adopting the above scheme, multi-modal sensors such as lidar and millimeter-wave radar set through FPGA hardware-level synchronization can effectively reduce data conflicts and time deviations; the designed improved Kalman filter combined with the weight allocation algorithm can effectively filter out burst noise (such as strong light interfering with camera data) and enhance the perception robustness in complex environments; it can dynamically adjust the noise variance prediction according to the sensor response speed, improving the real-time performance and accuracy of data fusion.

[0113] Among them, the adaptive flight control module includes an attitude controller based on deep reinforcement learning, and its reward function is:

[0114]

[0115] Among them:

[0116] k1, k2, k3 are dynamically adjusted according to the flight mode: the weights are (0.7, 0.2, 0.1) in the cruise stage and (0.4, 0.1, 0.5) in the obstacle avoidance stage;

[0117] Suppress control instruction jitter, where \(0.01\leqslant\gamma\leqslant0.1\), \(V\) is the current real-time flight speed of the UAV (unit: m / s), and \(V_{max}\) is the maximum designed flight speed of the UAV (unit: m / s), which is comprehensively determined by the power system, structural strength and aerodynamic performance;

[0118] P 电机( \(P(t)\): the real-time power of the motor, detected through current-voltage closed-loop;

[0119] Example:

[0120] When the UAV avoids obstacles, the reward function automatically adjusts the weight (\(k_3\) is increased to 0.5), giving priority to ensuring the obstacle avoidance success rate. When the flight speed increases, \(\gamma\) increases, smoothing the control instruction and reducing high-frequency jitter;

[0121] Technical effect: The control response delay is significantly reduced, and the obstacle avoidance success rate in complex environments is greatly improved.

[0122] Working principle process:

[0123] 1. The deep reinforcement learning model receives the flight state (attitude angle, speed, etc.).

[0124] 2. Calculate the action value according to the real-time reward function.

[0125] 3. Optimize the control strategy and output the servo instruction.

[0126] 4. The aerodynamic parameter self-compensation algorithm real-time corrects the aerodynamic coefficient matrix.

[0127] The adaptive flight control module also includes an aerodynamic parameter self-compensation algorithm, which real-time corrects the aerodynamic coefficient matrix through the wing surface pressure sensor.

[0128] Adopting the above scheme, through the deep reinforcement learning model combined with the dynamic reward function (tracking accuracy, energy consumption, obstacle avoidance success rate), intelligent decision-making in complex scenarios can be achieved; through the speed-adaptive smoothing coefficient (\(\gamma\)), high-frequency control jitter can be effectively suppressed, improving flight stability; through the self-compensation algorithm based on the wing surface pressure sensor, the aerodynamic coefficient can be dynamically adjusted to optimize flight attitude control.

[0129] Among them, the dynamic power management module includes: a multi-battery pack parallel topology structure and a battery health status evaluation unit:

[0130] The multi-battery pack parallel topology structure uses a bidirectional DC-DC converter to achieve dynamic energy allocation;

[0131] The battery health status evaluation unit calculates the aging factor based on electrochemical impedance spectroscopy:

[0132]

[0133] Where: SOH i is the state of health of the i-th battery pack (unitless, 0 ≤ SOH ≤ 1); C 实际 is the current actual available capacity of the battery pack (unit: Ah or mAh), reflecting the true energy storage capacity after battery aging; C 标称 is the rated capacity of the battery at the time of factory shipment (unit: Ah or mAh), serving as the reference capacity for calculating the state of health; μ is the aging attenuation coefficient, determined by battery materials and usage environment, and its value range is usually 0.001 - 0.01 (the specific value needs to be calibrated through experiments), representing the rate of capacity attenuation per cycle; N is the number of complete charge and discharge cycles experienced by the battery pack (unit: times), directly reflecting the cumulative aging degree of the battery;

[0134] The SOH formula is used for aging prediction: quantifying cycle aging through μ·N, combined with the actual capacity attenuation, to accurately evaluate the remaining battery life;

[0135] The objective function of the temperature - load joint optimization strategy of the dynamic power management module is extended to:

[0136]

[0137] Where:

[0138] α, β, η are weight coefficients, used to balance the priorities of temperature control, power distribution, and battery life, with dynamic adjustment rules:

[0139] High - temperature environment (such as 40°C): α = 0.6, β = 0.3, η = 0.1, giving priority to suppressing temperature rise;

[0140] Battery aging (such as SOH < 0.7): α = 0.2, β = 0.3, η = 0.5, focusing on extending battery life;

[0141] T max is the real - time maximum temperature of the battery pack (unit: °C), restricting the battery operating temperature to prevent thermal runaway;

[0142] T opt is the dynamic optimal operating temperature (unit: °C), with the calculation formula:

[0143] T opt = 25 + 0.1·(SOH i - 0.8)·100;

[0144] When SOH = 0.8, T opt = 25°C (ideal temperature);

[0145] When SOH decreases, the temperature is appropriately increased to compensate for capacity attenuation;

[0146] Pi is the actual output power of the i-th battery pack (unit: W);

[0147] P req is the power currently required by the system (unit: W);

[0148] P rated is the rated power of the battery pack (unit: W), that is, the maximum safe output power;

[0149] Example: In a high-temperature environment, the optimization goal focuses on temperature control (α = 0.6) to reduce the temperature of the battery pack; when the battery is aging (SOH i = 0.6), the weight η = 0.5 gives priority to extending the life.

[0150] Technical effect: The life of the battery pack is greatly extended, and the risk of thermal runaway is significantly reduced.

[0151] Working principle process:

[0152] 1. Multiple battery packs dynamically allocate energy through a bidirectional DC-DC converter.

[0153] 2. Real-time monitor the battery temperature, load demand and health status.

[0154] 3. Calculate the optimal power distribution strategy according to the objective function.

[0155] 4. Achieve μs-level power distribution through an intelligent power distribution matrix.

[0156] Adopting the above scheme, through the combination of electrochemical impedance spectroscopy analysis and aging factor calculation, the remaining life of the battery can be accurately evaluated, and the service cycle can be extended; through the temperature-load dynamic weight distribution strategy (focusing on temperature control at high temperature and life at aging) to balance safety and energy efficiency, the risk of thermal runaway can be effectively reduced; through the bidirectional DC-DC converter, it can support the flexible energy allocation of multiple battery packs to adapt to sudden load demands.

[0157] Among them, the distributed communication architecture module adopts an adaptive frequency hopping communication protocol, a quantum encryption transmission unit, and a fiber optic communication redundant link.

[0158] Among them, the adaptive frequency hopping communication protocol dynamically switches the frequency band according to the channel quality;

[0159] Among them, the quantum encryption transmission unit supports the key distribution of the BB84 protocol;

[0160] Among them, the fiber optic communication redundant link satisfies the mechanical constraint equation:

[0161]

[0162] Among them:

[0163] dL / dt is the optical fiber release rate, which is 0.5·V in the emergency obstacle avoidance mode max ;

[0164] θ bend ≤15°, the optical fiber bending angle, triggering the automatic retraction protection mechanism when exceeding 15°;

[0165] Vmax is the maximum designed flight speed of the drone (unit: m / s);

[0166] Example: When the drone flies at the maximum speed V max = 30 m / s, the optical fiber release rate is limited to 15 m / s. If the optical fiber bending angle reaches 20°, it will automatically retract and protect to avoid loss.

[0167] Technical effect: The optical fiber transmission loss is significantly reduced, and the reliability of the communication link is greatly enhanced.

[0168] Working principle process:

[0169] 1. The optical fiber reel dynamically releases the optical fiber according to the flight speed;

[0170] 2. Real-time monitor the bending angle, and trigger the protection mechanism when it exceeds the limit;

[0171] 3. The main control node switches to the optical fiber communication mode through the star topology network.

[0172] Adopting the above scheme, the anti-interception and anti-jamming capabilities of wireless communication can be effectively enhanced through the adaptive frequency hopping protocol and quantum encryption technology; through the optical fiber redundant link design combined with the bending angle limit (≤15°), the risk of communication interruption can be effectively reduced, ensuring high-bandwidth data transmission; through the real-time adaptation of the optical fiber release rate to the flight speed, mechanical loss can be effectively avoided.

[0173] Among them, the intelligent task decision-making module includes a multi-objective optimization decision-making engine, a task rehearsal subsystem based on digital twin, and a human-machine collaborative decision-making interface, where:

[0174] The multi-objective optimization decision-making engine uses the NSGA-II algorithm to balance task efficiency and energy consumption;

[0175] The task rehearsal subsystem based on digital twin simulates the flight environment through a high-precision dynamics model;

[0176] The human-machine collaborative decision-making interface supports natural language instruction parsing and multi-modal intention recognition.

[0177] Adopting the above solution, by using the NSGA-II multi-objective optimization engine to balance task efficiency and energy consumption, the success rate of task execution can be effectively improved; by using the high-precision dynamics model based on digital twin to simulate the real flight environment, the task risk can be effectively reduced; by simplifying the operation process through natural language parsing and multi-modal intention recognition, the decision-making flexibility can be effectively improved.

[0178] Among them, the fault self-repair system includes a dual-redundancy actuator, a key component health monitoring network, and a dynamic control system reconfiguration algorithm, where:

[0179] The dual-redundancy actuator adopts a heterogeneous drive design with electromagnetic actuators and hydraulic actuators in parallel;

[0180] The key component health monitoring network detects abnormal frequency components based on vibration spectrum analysis;

[0181] The dynamic control system reconfiguration algorithm maintains stability through torque redistribution when some propellers fail, satisfying:

[0182] Where:

[0183] F i is the lift force of the i-th propeller (unit: N), representing the vertical force generated by the propeller, which directly affects the lift and attitude control of the UAV;

[0184] R i is the arm length of the i-th propeller to the center of gravity of the UAV (unit: m), representing the influence degree of the lift force on the rotation of the airframe. The longer the arm length, the greater the torque;

[0185] I is the moment of inertia of the airframe (unit: kg·m 2 ), which characterizes the inertia of the airframe around the rotation axis and reflects the ability of the UAV to resist changes in angular acceleration;

[0186] θ is the angular acceleration of the airframe (unit: rad / s 2 ), representing the acceleration of the UAV rotating around the center of gravity, which needs to be controlled by torque balance;

[0187] Torque balance mechanism: When some propellers fail, the remaining propellers adjust the lift force F 1， to make the total torque (∑F i ·r i ) cancel the external interference torque (such as wind force or center of gravity offset) and maintain I·θ = 0 (stable state);

[0188] Dynamic adjustment process:

[0189] Step 1: Detect the position and quantity of the failed propellers;

[0190] Step 2: Calculate the required lift force F based on the lever arm length r1 of the remaining propellers i ;

[0191] Step 3: Control the lift force F by adjusting the motor speed i to satisfy the moment balance equation;

[0192] Step 4: Monitor the angular acceleration θ in real time and correct the lift force distribution through closed-loop feedback

[0193] Technical effect: The fault recovery time is significantly shortened, and the attitude deviation is small when a single propeller fails

[0194] Adopting the above technical solution, dual-backup drive is provided through heterogeneous actuators (electromagnetic + hydraulic servo), effectively reducing the risk of system failure caused by a single fault; through the coordinated adjustment of the propeller lift force and lever arm, the flight stability is quickly restored, effectively reducing the attitude deviation; through the vibration spectrum analysis, the anomalies of key components are diagnosed in real time, and potential faults can be accurately predicted in advance

[0195] Among them, the system is configured with an anti-interference navigation and positioning module, including:

[0196] Multi-mode satellite positioning receiver (GPS / Beidou / Galileo);

[0197] Visual SLAM positioning unit, based on the ORB feature matching algorithm;

[0198] Abnormal data cross-verification unit, using Mahalanobis distance detection:

[0199]

[0200] When D M > 3σ, data correction is triggered;

[0201] Among them: D M is the Mahalanobis distance; x is the current sensor data vector (a multi-dimensional vector composed of GPS coordinates, speed, etc.); μ is the mean vector of historical data, representing the expected value under normal conditions; Σ is the covariance matrix, describing the correlation and variance between sensor data; T is the transpose operator, ensuring the legality of vector and matrix multiplication;

[0202] Example: When the GPS positioning data and the visual SLAM data are quite different, calculate the Mahalanobis distance D M , if D M > 3σ, data correction is triggered, and the signal source with higher confidence is selected

[0203] Technical effect: The positioning error is significantly reduced, and the detection accuracy of abnormal data is greatly improved

[0204] Working principle process:

[0205] 1. Multi - mode positioning data input cross - validation unit;

[0206] 2. Calculate the Mahalanobis distance to judge data anomalies;

[0207] 3. If D M > 3σ, trigger the calibration algorithm to fuse the trusted data sources.

[0208] Adopting the above - mentioned scheme, by complementing satellite positioning (GPS / Beidou) with visual SLAM, the positioning accuracy in complex environments can be effectively improved; through the cross - validation mechanism based on the Mahalanobis distance, abnormal sensor data can be accurately identified and eliminated, enhancing the navigation reliability; through multi - mode redundant design, the impact of a single signal source failure on positioning can be effectively reduced.

[0209] Among them, the system adopts a modular and scalable architecture, including:

[0210] A hot - swappable hardware interface that complies with the ARINC810 standard;

[0211] An open - source software protocol stack that provides API interfaces for third - party algorithm integration;

[0212] The ability of in - air software upgrade, which realizes the incremental update of firmware through differential compression technology, and the compression ratio ≥ 70%.

[0213] Adopting the above - mentioned scheme, through the hot - swappable hardware interface and the open - source software protocol stack, rapid function expansion and third - party integration can be supported; through differential compression technology to achieve incremental firmware update, the bandwidth occupancy and upgrade time can be effectively reduced; through modular design, the replacement of faulty components is simplified, and the operation and maintenance complexity is reduced.

[0214] Among them, the application method in extreme failure scenarios:

[0215] When three propellers fail are detected, trigger the passive disaster - tolerant control algorithm, and generate equivalent lift through the periodic variable speed of the remaining propellers, and satisfy the angular momentum conservation equation:

[0216] Realize a safe return.

[0217] Among them, J is the moment of inertia of the fuselage, which characterizes the inertia of the UAV around the rotation axis and reflects its ability to resist changes in angular acceleration;

[0218] w is the instantaneous angular velocity of the fuselage (unit: rad / s), which describes the speed and direction of the UAV rotating around the center of gravity;

[0219] ΔL iΔL is the lift increment of the i-th remaining propeller (unit: N), which is generated by periodic variable speed. By adjusting the propeller speed, the lift is changed to offset the unbalanced moment. The calculation method is as follows:

[0220] where k is the lift coefficient, and Δn i is the change in rotational speed;

[0221] r i is the arm length of the i-th remaining propeller to the center of gravity of the UAV (unit: m), which is the moment contribution of the lift increment to the rotation of the fuselage;

[0222] k is the number of remaining available propellers.

[0223] The UAV generates angular momentum imbalance (J·w) due to the failure of some propellers; the remaining propellers generate a reverse moment (∑ΔL i ×r i ) through the lift increment ΔL i to make the total angular momentum zero and maintain stability.

[0224] For example, when three propellers fail, the remaining propellers have periodic variable speed (ΔL i changes) to offset the yaw moment (J·w) and achieve stable return.

[0225] Technical effect: It supports stable control under the failure of three propellers. The traditional scheme can only handle the failure of a single propeller. When three propellers fail, the flight stability can still be maintained, and the success rate of return is greatly improved.

[0226] Working principle process:

[0227] 1. The fault self-repair system detects the failure of the propeller;

[0228] 2. The passive disaster tolerance algorithm calculates the lift requirement of the remaining propellers;

[0229] 3. Adjust the rotational speed to satisfy the conservation of angular momentum and control the return path.

[0230] Through the periodic variable speed regulation of the remaining propellers, the angular momentum balance is still maintained when three propellers fail, and stable return is achieved. By the coordinated adjustment of the lift increment and the arm length, the unbalanced moment is offset, and the out-of-control of the fuselage is avoided. Breaking through the limitations of traditional redundant design, it supports safe operation in the scenario of multiple propeller failures. By dynamically adjusting the coordinated action of the lift (ΔL i ) of the remaining propellers and the arm length (r i ), combined with the inertia (J) and rotational state (w) of the fuselage, this solution can still maintain the conservation of angular momentum under extreme faults and ensure safe return.

[0231] In summary, the large UAV intelligent flight control integration system provided in this embodiment has the following advantages:

[0232] 1. Enhanced Environmental Perception and Fusion Capability

[0233] Multi-modal sensors (lidar, millimeter-wave radar, visible light camera, infrared sensor) significantly improve the target recognition accuracy and anti-interference ability in complex environments through hardware-level synchronous fusion and dynamic weight allocation algorithms; the real-time error compensation system suppresses burst noise to ensure the long-term stability of the perception data.

[0234] 2. Optimized Intelligent Flight Control

[0235] The deep reinforcement learning model combines dynamic reward functions (tracking accuracy, energy consumption, obstacle avoidance success rate) to achieve adaptive decision-making in complex scenarios; the real-time self-compensation of aerodynamic parameters optimizes the flight attitude, enhancing the stability and safety of high-maneuver tasks.

[0236] 3. Efficient Energy Management

[0237] The dynamic allocation strategy of multiple battery packs combined with temperature-load joint optimization balances energy consumption and safety, extends battery life and reduces the risk of thermal runaway; the intelligent power distribution matrix supports μs-level power distribution to adapt to sudden load demands.

[0238] 4. Reliable Communication and Navigation

[0239] Quantum encryption and fiber optic redundant link design improve the anti-interference ability of communication, ensuring the continuity of high-bandwidth data transmission; multi-mode positioning and cross-verification of abnormal data enhance the navigation accuracy and reliability in complex environments.

[0240] 5. Multi-level Fault Tolerance

[0241] Heterogeneous redundant actuators and dynamic torque distribution algorithms quickly restore flight stability, supporting safe return in scenarios where multiple propellers fail; the health monitoring network diagnoses key component abnormalities in real time and gives early warnings of potential faults.

[0242] 6. Modular Expansion and Convenient Maintenance

[0243] Hot-swap hardware interfaces and open software protocol stacks support rapid function expansion and third-party integration; the in-air software upgrade ability reduces maintenance costs and improves the system iteration efficiency.

[0244] The working process of this solution is as follows:

[0245] S1: Environmental Perception and Data Fusion

[0246] Data acquisition: Multi-modal sensors synchronously collect environmental information (obstacles, terrain, meteorology, etc.).

[0247] Error Compensation: Based on the improved Kalman filter and dynamic weight allocation algorithm, noise is filtered and data is fused.

[0248] Transmission Link: The sensed data is transmitted in real time to the intelligent task decision-making module through a fiber optic redundant link.

[0249] S2: Task Planning and Decision Making

[0250] Multi-objective Optimization: The NSGA-II algorithm balances task efficiency, energy consumption and safety to generate the optimal flight path.

[0251] Digital Twin Rehearsal: The flight environment is simulated through a high-precision dynamics model to verify the feasibility of the task.

[0252] Human-Machine Collaboration: Parse natural language instructions and adjust task strategies by combining multi-modal intent recognition.

[0253] S3: Adaptive Flight Control

[0254] Attitude Regulation: The deep reinforcement learning model outputs control instructions according to the real-time reward function to optimize the flight trajectory.

[0255] Aerodynamic Compensation: The wing surface pressure sensors dynamically correct the aerodynamic coefficient matrix to adapt to airflow disturbances.

[0256] Fault Response: After detecting the failure of the propeller or servo, trigger the dynamic torque distribution algorithm to restore balance.

[0257] S4: Energy Management and Communication Assurance

[0258] Power Allocation: The intelligent power distribution matrix dynamically allocates the energy of multiple battery packs to give priority to meeting the needs of high-load modules.

[0259] Communication Redundancy: The adaptive frequency hopping protocol and the fiber optic link work together to ensure the continuity and security of data transmission.

[0260] S5: Fault Diagnosis and Disaster Tolerance Control

[0261] Health Monitoring: Vibration spectrum analysis detects abnormalities in the actuator, triggering early warnings or redundant switching.

[0262] Extreme Disaster Tolerance: When three propellers fail, the remaining propellers periodically change their rotational speeds to generate equivalent lift, maintaining angular momentum conservation and returning safely.

[0263] The above are only the preferred embodiments of the present invention, and do not limit the implementation manners and protection scope of the present invention. For those skilled in the art, it should be able to realize that all equivalent replacements and obvious changes made by using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent flight control integration system for large unmanned aerial vehicles, characterized in that, include: Multimodal environmental perception module, adaptive flight control module, dynamic power management module, distributed communication architecture module, intelligent mission decision module and fault self-repair system; The multimodal environment perception module is connected to the adaptive flight control module via a bidirectional real-time data bus and transmits environment perception data to the intelligent task decision module via a redundant optical fiber communication link; The dynamic power management module is connected to each functional module through an intelligent power distribution matrix, which includes a 16-channel programmable MOSFET array and supports μs-level dynamic power distribution; The distributed communication architecture module connects the master node and at least three redundant sub-nodes through a star topology network, and a time division multiplexing + frequency division multiplexing hybrid communication protocol is adopted between the nodes; The fault self-repair system is interconnected with all actuators via a dual-redundancy CAN bus and is embedded in the load monitoring interface of the dynamic power management module to achieve current-temperature joint diagnosis.

2. The large unmanned aerial vehicle intelligent flight control integration system according to claim 1, characterized in that, The multimodal environment perception module includes: The data of the laser radar unit, millimeter-wave radar unit, visible light camera unit and infrared sensor unit are synchronized and integrated at the hardware level through the FPGA chip; Equipped with a real-time error compensation system based on an improved Kalman filter, its dynamic weight distribution function is: where: λ i , λ j are both attenuation factors (value range 0.5 - 2); Δz i (t) is the measurement residual of the i-th sensor at time t; Δz j (t) is the measurement residual of the j-th sensor at time t; τ is the sensor response delay compensation (0≤τ≤50ms); is the noise variance of sensor i after delay compensation, and dynamically predicts the noise level at the future τ moment.

3. The large unmanned aerial vehicle intelligent flight control integration system according to claim 2, characterized in that The adaptive flight control module includes a deep reinforcement learning-based attitude controller, whose reward function is: Wherein: k1, k2, k3 are dynamically adjusted according to the flight mode: the weights in the cruise stage are (0.7, 0.2, 0.1), and in the obstacle avoidance stage are (0.4, 0.1, 0.5); Suppress the jitter of the control command, 0.01 ≤ γ ≤ 0.1; V is the current real-time flight speed of the UAV (unit: m / s); Vmax is the maximum designed flight speed of the UAV (unit: m / s); P 电机 ( t): the real-time power of the motor, detected through the current-voltage closed loop.

4. The large unmanned aerial vehicle intelligent flight control integration system according to claim 3, wherein The dynamic power management module includes: Multi-battery parallel topology, using bidirectional DC-DC converter to achieve dynamic energy allocation; Battery health status assessment unit, which calculates the aging factor based on electrochemical impedance spectroscopy: Where: SOH i is the state of health of the i-th battery pack; C 实际 is the current actual available capacity of the battery pack (unit: Ah or mAh); C 标称 is the rated capacity at the time of battery factory (unit: Ah or mAh); μ is the aging attenuation coefficient; N is the number of complete charge and discharge cycles experienced by the battery pack (unit: times); The temperature-load joint optimization strategy objective function of the dynamic power management module is expanded to: Where: α, β, η are weight coefficients; T max is the real-time maximum temperature of the battery pack (unit: °C); T opt is the dynamic optimal operating temperature (unit: °C); P i is the actual output power of the i-th battery pack (unit: W); P req is the power currently required by the system (unit: W); P rated is the rated power of the battery pack (unit: W).

5. The large unmanned aerial vehicle intelligent flight control integration system according to claim 1, wherein, The distributed communication architecture module adopts: Adaptive frequency hopping communication protocol, dynamically switching frequency bands based on channel quality; Quantum encryption transmission unit, supporting BB84 protocol key distribution; Fiber optic communication redundant links satisfy the mechanical constraint equation: where: dL / dt is the optical fiber release rate, which is 0.5·V in the emergency obstacle avoidance mode max ; θ bend ≤15°, the optical fiber bending angle, and when it exceeds 15°, the automatic retraction protection mechanism is triggered; Vmax is the maximum designed flight speed of the drone (unit: m / s).

6. The large-scale unmanned aerial vehicle intelligent flight control integrated system according to claim 5 is characterized in that: The intelligent task decision module includes: Multi-objective optimization decision engine, using NSGA-II algorithm to balance task efficiency and energy consumption; The mission rehearsal subsystem based on digital twin simulates the flight environment through a high-precision dynamic model; Human-computer collaborative decision-making interface supports natural language instruction parsing and multimodal intent recognition.

7. The intelligent flight control integration system for large unmanned aerial vehicles according to claim 1, characterized in that The fault self-repair system includes: Dual redundant actuators, using a heterogeneous drive design with electromagnetic servos and hydraulic servos in parallel; Key component health monitoring network, detecting abnormal frequency components based on vibration spectrum analysis; A dynamic control system reconstruction algorithm that maintains stability through torque redistribution when some propellers fail and satisfies: Where: F i is the lift force of the i-th propeller (unit: N); R i is the length of the lever arm from the i-th propeller to the center of gravity of the UAV (unit: m); I is the moment of inertia of the airframe (unit: kg·m 2 ); θ is the angular acceleration of the airframe (unit: rad / s 2 ).

8. The large unmanned aerial vehicle intelligent flight control integration system according to claim 1, characterized in that The system is equipped with an anti-interference navigation and positioning module, including: Multi-mode satellite positioning receiver; Visual SLAM positioning unit, based on ORB feature matching algorithm; Abnormal data cross-validation unit, using Mahalanobis distance detection: When D M > 3σ, data correction is triggered; Where: D M is the Mahalanobis distance; x is the current sensor data vector; μ is the mean vector of historical data; Σ is the covariance matrix; T is the transpose operator.

9. The intelligent flight control integration system for large unmanned aerial vehicles according to claim 1, characterized in that, The system adopts a modular and scalable architecture, including: Support hot-swappable hardware interface, compliant with ARINC810 standard; Open software protocol stack, providing API interface for third-party algorithm integration; The ability of over-the-air software upgrade, which realizes the incremental update of firmware through differential compression technology, with a compression rate ≥ 70%.

10. The large unmanned aerial vehicle intelligent flight control integration system according to claim 1, characterized in that Application method in extreme failure scenarios: When three propellers are detected to fail, trigger the passive disaster tolerance control algorithm, generate equivalent lift through the periodic speed change of the remaining propellers, and satisfy the angular momentum conservation equation: Realize a safe return.

Images