Multi-rotor unmanned aerial vehicle paddle fault autonomous control system

Abstract

The invention discloses a multi-rotor unmanned aerial vehicle blade fault autonomous control system, which comprises a multi-sensor fusion module and an intelligent algorithm optimization module, and is characterized in that the multi-sensor fusion module is used for acquiring acceleration, angular velocity, strain and visual image data of blades in real time and performing data fusion processing through a Kalman filtering algorithm so as to improve the accuracy and reliability of the data. According to the multi-rotor unmanned aerial vehicle blade fault autonomous control system, through multi-sensor fusion and a Kalman filtering algorithm, noise interference is effectively filtered out, and the data accuracy is improved. The deep learning algorithm can automatically learn blade fault features, high-precision fault detection is realized, the misjudgment rate is reduced by more than 80%, the control strategy optimized by the reinforcement learning algorithm can output an optimal control instruction within 0.1 second after blade faults occur, the flight attitude of the unmanned aerial vehicle is adjusted, and the flight safety is ensured. Environment parameters can be monitored in real time through the environment sensing and self-adaption module, and the system can stably operate in the temperature range of-20 DEG C to 50 DEG C and the wind speed range of 0-10 m / s.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-rotor unmanned aerial vehicles, specifically to a multi-rotor unmanned aerial vehicle propeller fault autonomous control system. BACKGROUND

[0002] Multi-rotor unmanned aerial vehicles are widely used in aerial photography, logistics, agriculture, inspection, and other fields. Their flight stability and safety largely depend on the normal operation of the propellers. Propeller faults can cause the unmanned aerial vehicle to lose balance, crash, and even cause personal injury and property loss.

[0003] In the prior art, the multi-rotor unmanned aerial vehicle propeller fault detection and control system has the following shortcomings:

[0004] Low fault detection accuracy: Most systems rely on a single type of sensor (such as an accelerometer or an angular speed meter), which is susceptible to external interference (such as airflow and electromagnetic interference), leading to misjudgment or missed judgment.

[0005] Slow response speed: The process from fault detection to executing control measures is complex and has high latency, making it impossible to respond to sudden faults in a timely manner.

[0006] Poor environmental adaptability: Different environmental conditions (such as temperature, humidity, and wind speed) have not been fully considered in the influence on propeller fault characteristics, resulting in a decline in system performance in complex environments.

[0007] Low system integration: The propeller fault detection system has low integration with the unmanned aerial vehicle flight control system and power system, and data exchange is not smooth, affecting overall performance.

[0008] In summary, the prior art has significant defects in fault detection accuracy, response speed, environmental adaptability, and system integration. To solve these problems, the present application proposes a multi-rotor unmanned aerial vehicle propeller fault autonomous control system that realizes rapid and accurate detection and autonomous control of propeller faults through multi-sensor fusion, intelligent algorithm optimization, environmental perception and self-adaptation, and system integration optimization. SUMMARY

[0009] (I) Technical problems solved

[0010] To address the shortcomings of the prior art, the present application provides a multi-rotor unmanned aerial vehicle propeller fault autonomous control system that has the advantages of rapid and accurate detection and autonomous control of propeller faults, and solves the problems of poor fault detection accuracy, response speed, environmental adaptability, and system integration.

[0011] (II) Technical solutions

[0012] To achieve the above-mentioned purposes, the present application provides the following technical solutions: a multi-rotor unmanned aerial vehicle propeller fault autonomous control system, comprising:

[0013] 1. Multi-sensor fusion module

[0014] 1.1 Sensor Configuration: Accelerometers, angular velocity sensors, and strain sensors are installed on each propeller blade, and visual sensors are placed at key locations on the drone fuselage. Accelerometers monitor blade vibration acceleration, angular velocity sensors monitor blade rotational angular velocity, strain sensors monitor blade root strain, and visual sensors capture images of the blade surface to assist in determining whether the blade has any external faults, such as cracks or breakage.

[0015] 1.2 Data acquisition and synchronization: The data of each sensor is collected synchronously through the data acquisition unit, and the data is synchronized using timestamps to ensure that the data is in the same time coordinate system;

[0016] 1.3 Data preprocessing: Filter and normalize the collected raw data to remove noise and outliers and improve data quality;

[0017] 1.4 Kalman filter algorithm:

[0018] State space model establishment: define the state vector of the blade

[0019] Among them, x k ,y k ,z k Represents the position of the blade in three directions, x k ,y k ,z k Represents the speed of the blade in three directions respectively;

[0020] Establish the state transfer equation: X k =FX k―1 +GW k―1

[0021] Among them, F is the state transfer matrix, G is the control input matrix, W k―1 is the process noise, and at the same time, the observation vector is defined as: Z k =[z x,k ,z y,k ,z z,k ] T

[0022] Establish the observation equation Z k =HX k +V k , where H is the observation matrix, V k is the observation noise;

[0023] Kalman filter iteration calculation: estimate the real state feature vector of the blade through the prediction step and the update step of Kalman filter. The prediction step includes state prediction and error covariance prediction where is the state estimation of the last time, is the error covariance estimation of the last time, and Q is the process noise covariance matrix. The update step includes Kalman gain calculation, state update and error covariance update, and error covariance update where R is the observation noise covariance matrix, and I is the identity matrix.

[0024] II. Intelligent algorithm optimization module

[0025] Deep learning unit:

[0026] 2.1 Model structure: use convolutional neural network (CNN), including input layer, several convolutional layers, pooling layers and fully connected layers. The input layer receives the fused data feature vector, the convolutional layer uses different size convolutional kernels to extract features, the pooling layer is used for dimension reduction and reduces the amount of calculation, and the fully connected layer classifies the features;

[0027] 2.2 Training data set: use a data set containing 2000 normal data and 2000 fault data for training, covering the fault types of cracks, fractures, looseness and other faults that may occur in the blade. The data set is strictly screened and labeled to ensure its representativeness and accuracy;

[0028] 2.3 Data augmentation technology: in the training process, the original data is transformed by rotation, translation, scaling, etc. to expand the data set size and improve the generalization ability of the model.

[0029] 2.4 Training algorithm: use Adam optimization algorithm, learning rate is 0.001, batch size is 32, and training epoch is 100. Through cross-validation and early stopping strategy, prevent model overfitting, ensure its accuracy on test set is higher than 95%.

[0030] 2.5 Fault identification result: the model outputs the fault type and confidence, and the confidence threshold is set to 0.8. When the confidence is higher than the threshold, the system determines that the blade has a fault and outputs the fault type; when the confidence is lower than the threshold, the system issues a warning, indicating that there may be potential faults, which need to be further monitored and analyzed.

[0031] Reinforcement learning unit:

[0032] 2.6 Simulation Training Environment: Construct high-precision unmanned aerial vehicle dynamics model and fault model, covering single-blade failure, multi-blade failure, different load conditions, different flight attitude and other complex scenarios.

[0033] 2.7 Reward Function Design: Define reward function

[0034] R(s,a)=w1×stability index+w2×safety index+w3×energy consumption index

[0035] where w1, w2, w3 are weight coefficients, representing the importance of flight stability, safety and energy consumption in the reward function,

[0036] Real-time control instruction generation: When detecting blade failure, according to the fault type and severity, output optimal control instruction, including adjusting the speed of other blades, changing flight attitude, starting safety landing program, etc. For example, when a single blade appears crack failure, the system can detect the fault within 0.1 seconds, and through adjusting the speed of other blades and flight attitude, the unmanned aerial vehicle maintains balanced flight, while smoothly landing at a deceleration rate of 1.5m / s 2 During landing, the system monitors the flight state of the unmanned aerial vehicle in real time, dynamically adjusts the landing speed according to the height and speed information, and ensures safe landing.

[0037] III. Environment Perception and Adaptive Module

[0038] Environment Information Collection

[0039] 3.1 Weather Sensor: Monitor temperature, humidity, air pressure, wind speed and direction, sampling frequency is 1Hz. High-precision weather station module is used, installed on the upper part of the unmanned aerial vehicle body, avoiding the interference of blade airflow. Temperature measurement range is -40℃ to 70℃, accuracy is better than ±0.5℃; humidity measurement range is 0% to 100%RH, accuracy is better than ±3%RH; air pressure measurement range is 300hPa to 1100hPa, accuracy is better than ±1hPa; wind speed measurement range is 0 to 30m / s, accuracy is better than ±0.5m / s; wind direction measurement range is 0° to 360°, accuracy is better than ±5°.

[0040] 3.2 Terrain Mapping Sensor: Obtain terrain height and obstacle information, update frequency is 2Hz. Laser radar or millimeter wave radar is used, installed at the bottom of the unmanned aerial vehicle, emitting detection signals downward, measuring the distance between the unmanned aerial vehicle and the ground in real time, and constructing terrain point cloud map. Measurement range is 0.1m to 100m, accuracy is better than ±0.1m.

[0041] 3.3 Electromagnetic Interference Sensor: Detects electromagnetic field intensity with a sampling frequency of 10 Hz. A wideband electromagnetic field intensity detection module is installed near the electronic equipment compartment of the UAV body to monitor electromagnetic interference sources in the flight environment. The measurement frequency range is 100 kHz to 3 GHz, the measurement range is 0 V / m to 100 V / m, and the accuracy is better than ±2 dB.

[0042] Adaptive Control Unit

[0043] 3.4 Environmental Parameter Analysis: Real-time analysis of collected environmental parameters to determine the complexity of the current flight environment and its impact on blade fault detection. For example, low temperature may cause blade material to become brittle, increasing the risk of crack failure; strong wind may cause blade vibration to intensify, interfering with fault detection.

[0044] 3.5 Parameter Adjustment Strategy: Automatically adjust the parameters and thresholds of the fault diagnosis algorithm based on the results of environmental parameter analysis. In low temperature environments, appropriately lower the threshold of crack fault diagnosis to improve detection sensitivity; in strong wind environments, increase the filtering strength of vibration fault diagnosis to reduce the false positive rate.

[0045] 3.6 Real-time Feedback and Optimization: Real-time feedback of adjusted parameters to the intelligent algorithm optimization module, and continuous optimization of adjustment strategies based on fault diagnosis results to ensure the best performance of the system under different environmental conditions.

[0046] Four, System Integration Optimization Module

[0047] 4.1 Hardware Interface Unit: Standardized hardware interface is adopted to ensure seamless connection between the blade fault self-control system and the flight control system, power system, communication system and other hardware devices of the UAV, improving the integration and compatibility of the system.

[0048] 4.2 Communication Protocol Unit: CAN bus or serial communication protocol is adopted to establish communication connection between sensors and control units, and between various systems, ensuring the stability and anti-interference ability of data transmission, and the data transmission delay does not exceed 50 ms to ensure the rapid execution of control commands.

[0049] 4.3 Software Integration Unit: Deeply integrate fault diagnosis software with flight control software and ground station software. Integrate fault diagnosis function in flight control software to enable flight control system to adjust flight control strategy in time according to fault diagnosis results; integrate fault information display and decision support function in ground station software to provide intuitive visual interface for operators, display UAV flight trajectory, attitude change and fault state in graphical way, and automatically provide decision suggestions such as continue flying, land nearby, return, etc. according to received fault information and flight state information.

[0050] Five, Fault Control Unit

[0051] 5.1 Control strategy development: According to the type and severity of the fault, combined with the current flight state and environmental information, the corresponding control strategy is developed. For example, in the case of single blade failure, by adjusting the rotation speed of other blades and flight attitude, the UAV can maintain balanced flight and safe landing; in the case of multiple blade failure or other serious faults, the emergency landing program is prioritized to ensure the UAV lands in a safe area and avoids causing greater losses.

[0052] 5.2 Flight attitude adjustment: According to the developed control strategy, the flight attitude of the UAV is adjusted in real time. By controlling the pitch angle, roll angle and heading angle of the UAV, the UAV can maintain stable flight in the event of a fault and avoid losing control. At the same time, according to the fault condition and environmental factors, the flight speed and height are dynamically adjusted to adapt to different flight conditions and safety requirements.

[0053] 5.3 Safe landing execution: After the fault occurs, according to the severity of the fault and the flight state, the safe landing program is started in time. During the landing process, the system monitors the flight state of the UAV in real time, dynamically adjusts the landing speed according to the height and speed information, and ensures safe landing. If new faults or abnormal conditions occur during the landing process, the system can make timely adjustments and take appropriate measures, such as hovering and waiting, re-planning the landing path, etc., to maximize the safety of the UAV and the surrounding environment.

[0054] Six, information feedback unit

[0055] 6.1 Fault information transmission: The fault type, location and severity information are transmitted to the ground station through the wireless communication module. Wireless communication uses Wi-Fi or 4G / 5G network, which automatically switches according to flight distance and environmental conditions. Data transmission uses encryption protocol to ensure information security. Fault information is sent in a structured data format, including fault code, fault description, occurrence time, location coordinates and other detailed content.

[0056] 6.2 Flight state information display: Real-time collection of UAV flight state information, including position (latitude and longitude, height), speed (horizontal speed, vertical speed), attitude (pitch angle, roll angle, heading angle), etc., with an update frequency of 10Hz. The ground station software displays the flight trajectory, attitude change and fault state of the UAV in a graphical manner, and has data recording and playback functions, which facilitates operators to analyze and troubleshoot after the event.

[0057] 6.3 Operator decision support: The ground station software automatically provides decision recommendations such as continuing flight, landing nearby, returning, etc. based on the received fault information and flight state information. The operator can remotely control the UAV through manual control or sending instructions according to the actual situation and software recommendations to ensure flight safety.

[0058] (III) Advantages

[0059] Compared with the prior art, the present application provides a multi-rotor unmanned aerial vehicle blade fault autonomous control system, which has the following advantages:

[0060] 1. The multi-rotor unmanned aerial vehicle blade fault autonomous control system effectively filters out noise interference and improves data accuracy through multi-sensor fusion and Kalman filtering algorithm. Deep learning algorithm can automatically learn blade fault features to realize high-precision fault detection, and the misjudgment rate is reduced by more than 80%. The control strategy optimized by reinforcement learning algorithm can output optimal control instructions within 0.1 seconds after the blade fault occurs, adjust the flight attitude of the unmanned aerial vehicle, and ensure flight safety.

[0061] 2. The multi-rotor unmanned aerial vehicle blade fault autonomous control system can monitor environmental parameters in real time through the environment perception and adaptive module, and dynamically adjust the fault diagnosis algorithm, so that the system can stably operate in the temperature range of-20℃ to 50℃ and the wind speed range of 0 to 10m / s. The standardization design of hardware interface and communication protocol improves the integration degree of the system by 60%, reduces the data transmission delay to within 30ms, and significantly improves the overall performance of the unmanned aerial vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 The system initialization flowchart proposed by the present application is shown in the figure;

[0063] Figure 2 The real-time monitoring flowchart proposed by the present application is shown in the figure;

[0064] Figure 3 The environment perception flowchart proposed by the present application is shown in the figure;

[0065] Figure 4 The fault diagnosis flowchart proposed by the present application is shown in the figure;

[0066] Figure 5 The adaptive adjustment flowchart proposed by the present application is shown in the figure;

[0067] Figure 6 The fault control and information feedback flowchart proposed by the present application is shown in the figure. DETAILED DESCRIPTION

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments 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 creative efforts are within the scope of protection of the present invention.

[0069] After the drone is turned on, each sensor performs a self-check, initializes the data acquisition and transmission modules, and loads the intelligent algorithm model into the control unit, ready to start working. The system initialization time does not exceed 5 seconds. The specific steps are as follows:

[0070] Sensor self-check: The system checks the working status of the acceleration sensor, angular velocity sensor, strain sensor, and visual sensor in sequence to ensure their normal operation. If a faulty sensor is found, the system will issue an alarm and record the fault information.

[0071] Initialize the data acquisition module: Set the data acquisition frequency and transmission parameters for each sensor. Set the acquisition frequency of the accelerometer and angular velocity sensor to 200Hz, the acquisition frequency of the strain sensor to 100Hz, and the frame rate of the visual sensor to 30fps.

[0072] Data transmission module initialization: establish the communication connection between the sensor and the control unit, use the CAN bus protocol for data transmission, and ensure the stability and anti-interference ability of data transmission.

[0073] Intelligent algorithm model loading: Pre-trained deep learning models and reinforcement learning models are loaded into the control unit's memory and initialized. The deep learning model is used for fault diagnosis, and the reinforcement learning model is used for post-fault control strategy optimization.

[0074] Initialize the environmental perception module: Start the meteorological sensor, terrain mapping sensor, and electromagnetic interference sensor, and set their sampling frequencies and operating modes. The meteorological sensor's sampling frequency is 1 Hz, the terrain mapping sensor's update frequency is 2 Hz, and the electromagnetic interference sensor's sampling frequency is 10 Hz.

[0075] Real-time monitoring

[0076] The multi-sensor fusion module collects the operating status data of the blades in real time:

[0077] Accelerometer: This sensor collects blade vibration acceleration at a frequency of 200 Hz, monitoring vibration amplitude and frequency changes during operation. Installed at the root of the blade, it uses a high-precision MEMS accelerometer with a measurement range of ±16g and a resolution better than 0.01g.

[0078] Angular velocity sensor: Collects the angular velocity of the blade rotation at a frequency of 200 Hz to monitor its rotational stability. The sensor is installed on the motor shaft and uses a fiber-optic gyroscope with a measurement range of ±1000° / s and an accuracy better than 0.1° / s.

[0079] Strain sensor: Real-time monitoring of the strain at the root of the blade to detect the bending and torsional stresses it experiences during flight. The sensor uses high-precision resistance strain gauges attached to specific locations on the blade root with a measurement range of ±2000με and an accuracy better than 1με.

[0080] Visual sensor: Captures images of the blade at a frame rate of 30 fps to identify visual defects such as cracks and damage on the blade surface through image processing algorithms. The sensor is installed at key locations on the UAV body and uses a high-resolution camera with optical image stabilization, with an image resolution of no less than 1280×720 pixels.

[0081] Data fusion: The data is processed through the Kalman filter algorithm to output the blade state feature vector. The specific steps are as follows:

[0082] Data synchronization: The data acquisition unit synchronizes the sensor data through timestamps to ensure that the data is in the same time coordinate system.

[0083] Data preprocessing: The raw data collected is filtered and normalized to remove noise and outliers, improving data quality.

[0084] Kalman filtering: The extended Kalman filter algorithm is used to fuse the preprocessed data to establish a spatial model and an observation model of the blade state, and through iterative calculation to estimate the true state feature vector of the blade. The state feature vector includes information such as blade vibration amplitude, rotational speed, strain value, and visual integrity.

[0085] Environmental perception

[0086] Real-time collection of environmental perception sensors:

[0087] Weather sensor: Monitors temperature, humidity, air pressure, wind speed, and wind direction with a sampling frequency of 1 Hz. The sensor uses a high-precision weather station module installed on the upper part of the UAV body to avoid interference from the blade airflow. The temperature measurement range is -40°C to 70°C with an accuracy better than ±0.5°C; the humidity measurement range is 0% to 100% RH with an accuracy better than ±3% RH; the air pressure measurement range is 300hPa to 1100hPa with an accuracy better than ±1hPa; the wind speed measurement range is 0 to 30m / s with an accuracy better than ±0.5m / s; the wind direction measurement range is 0° to 360° with an accuracy better than ±5°.

[0088] Terrain mapping sensor: Acquire terrain height and obstacle information, update frequency is 2Hz. The sensor uses laser radar or millimeter wave radar, installed at the bottom of the UAV, emits detection signals downward, measures the distance between the UAV and the ground in real time, and constructs a terrain point cloud map. The measurement range is 0.1m to 100m, and the accuracy is better than ±0.1m.

[0089] Electromagnetic interference sensor: Detect electromagnetic field intensity, sampling frequency is 10Hz. The sensor uses a wideband electromagnetic field intensity detection module, installed near the electronic equipment compartment of the UAV body, monitors electromagnetic interference sources in the flight environment. The measurement frequency range is 100kHz to 3GHz, the measurement range is 0V / m to 100V / m, and the accuracy is better than ±2dB.

[0090] Environmental information is used to dynamically adjust fault diagnosis algorithm parameters. The specific steps are as follows:

[0091] Environmental parameter analysis: The adaptive control unit analyzes the collected environmental parameters in real time, judges the complexity of the current flight environment and the influence degree on blade fault detection. For example, low temperature may cause blade material to become brittle, increasing the risk of crack failure; strong wind may cause blade vibration to intensify, interfering with fault detection.

[0092] Parameter adjustment strategy: According to the results of environmental parameter analysis, the adaptive control unit automatically adjusts the parameters and thresholds of the fault diagnosis algorithm. In a low-temperature environment, appropriately lower the threshold of crack fault diagnosis to improve detection sensitivity; in a strong wind environment, increase the filtering strength of vibration fault diagnosis to reduce the false positive rate.

[0093] Real-time feedback and optimization: The adaptive control unit feeds back the adjusted parameters to the intelligent algorithm optimization module in real time, and continuously optimizes the adjustment strategy according to the fault diagnosis results to ensure the best performance of the system under different environmental conditions.

[0094] Fault diagnosis

[0095] Intelligent algorithm optimization module:

[0096] Deep learning unit: Receives the fused data feature vector, and performs fault recognition through the trained CNN model. The CNN model uses convolutional layers, pooling layers and fully connected layers to automatically extract and classify the fused data features. The model training data set contains 2000 groups of normal data and 2000 groups of fault data, covering crack, fracture, loosening and other fault types. The data set has undergone strict screening and labeling to ensure its representativeness and accuracy.

[0097] Model training process: Data augmentation techniques are used to rotate, translate, and scale the original data to expand the dataset size and improve the model's generalization ability. The Adam optimization algorithm is used for training with a learning rate of 0.001, a batch size of 32, and a training epoch of 100. Cross-validation and early stopping strategies are used to prevent overfitting and ensure that the model's accuracy on the test set is greater than 95%.

[0098] Fault identification results: Include fault type and confidence level, with a confidence threshold set at 0.8. When the confidence level is higher than the threshold, the system determines that the blade has failed and outputs the fault type; when the confidence level is lower than the threshold, the system issues a warning, indicating that there may be potential faults that need further monitoring and analysis.

[0099] Adaptive adjustment

[0100] Adaptive control unit:

[0101] Temperature compensation: Adjust the noise parameters of the sensor data fusion algorithm according to the change in environmental temperature. In low-temperature environments, sensor noise may increase, and the adaptive control unit increases the diagonal elements of the noise covariance matrix of the Kalman filter algorithm to improve the robustness of the filter to noise and ensure the accuracy of data fusion.

[0102] Wind speed compensation: Dynamically adjust the sensitivity of the fault diagnosis algorithm according to wind speed information. In strong wind environments, blade vibration increases, and the adaptive control unit reduces the threshold of the fault diagnosis algorithm to increase the sensitivity to vibration faults and avoid missing faults due to wind interference.

[0103] Electromagnetic interference suppression: In strong electromagnetic interference environments, increase the data filtering strength to ensure diagnostic accuracy. The adaptive control unit adjusts the cutoff frequency and filter order of the digital filter to enhance the suppression of electromagnetic interference and improve the quality of sensor data, ensuring the reliability of fault diagnosis.

[0104] Fault control

[0105] Reinforcement learning unit:

[0106] Simulation training: In a simulated flight environment, the agent learns the optimal control strategy in different fault scenarios. The simulation environment uses high-precision drone dynamics and fault models, covering single-blade faults, multi-blade faults, different load conditions, different flight attitudes, and other complex scenarios. The agent learns through interaction with the environment, trial and error, and experience accumulation, gradually mastering the best control strategy in various fault conditions.

[0107] Reward function design: Define the reward function to evaluate and guide the behavior of the agent. The reward function includes indicators such as flight stability, safety, and energy consumption of the UAV. For example, when the UAV maintains balanced flight and safely lands after a fault occurs, the agent is given a positive reward; when the UAV loses control or crashes, it is given a negative reward. By continuously optimizing the reward function, the agent learns a better control strategy.

[0108] Real-time control instruction generation: When a propeller fault is detected, the optimal control instruction is output according to the fault type and severity. Control instructions include adjusting the speed of other propellers, changing flight attitude, starting a safe landing program, etc. For example, when a single propeller has a crack fault, the system can detect the fault within 0.1 seconds and adjust the speed of other propellers and flight attitude to maintain balanced flight, while smoothly landing at a speed of 1.5 m / s 2 Real-time control instruction generation: When a propeller fault is detected, the optimal control instruction is output according to the fault type and severity. Control instructions include adjusting the speed of other propellers, changing flight attitude, starting a safe landing program, etc. For example, when a single propeller has a crack fault, the system can detect the fault within 0.1 seconds and adjust the speed of other propellers and flight attitude to maintain balanced flight, while smoothly landing at a speed of 1.5 m / s

[0109] Information feedback

[0110] The system feeds back fault information and flight state information of the UAV to the ground station in real time:

[0111] Fault information transmission: Fault type, location, and severity are transmitted to the ground station through the wireless communication module. Wireless communication uses Wi-Fi or 4G / 5G network, automatically switching according to flight distance and environmental conditions. Data transmission uses encryption protocol to ensure information security. Fault information is sent in a structured data format, including fault code, fault description, occurrence time, location coordinates, etc.

[0112] Flight state information display: Flight state information includes the position (latitude and longitude, height), speed (horizontal speed, vertical speed), attitude (pitch angle, roll angle, heading angle) of the UAV, with an update frequency of 10 Hz. The ground station software provides an intuitive visual interface to display the flight trajectory, attitude change, and fault state of the UAV in a graphical manner. At the same time, the software has data recording and playback functions, facilitating operators to analyze and troubleshoot after the fact.

[0113] Operator decision support: The ground station software automatically provides decision suggestions such as continuing to fly, landing nearby, returning, etc. based on the received fault information and flight state information. Operators can remotely control the UAV through manual control or sending instructions according to actual conditions and software suggestions to ensure flight safety.

[0114] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and implementations, it is to be understood that the terminology used is for the purpose of descriptive clarity and that it is intended to be limited only by the words recited in the appended claims. It is to be understood that the terms "including", "comprising", "consisting" and variations thereof do not preclude the addition of further integers to the combinations of integers specified in the claims. It is to be understood that the terms "including", "comprising", "consisting" and variations thereof encompass the various features of the embodiments described herein. It is not intended that the application be limited to the implementation that is described in detail and / or shown in the drawings. It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and implementations, it is to be understood that the terminology used is for the purpose of descriptive clarity and that it is intended to be limited only by the words recited in the appended claims.

[0115] While the embodiments of the application have been shown and described herein, it is to be understood that the application is not limited to the embodiments described, and it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and implementations, it is to be understood that the terminology used is for the purpose of descriptive clarity and that it is intended to be limited only by the words recited in the appended claims.

Claims

1. The autonomous control system for multi-rotor UAV blade failure is characterized by: include: a. Multi-sensor fusion module, used to collect blade acceleration, angular velocity, strain, and visual image data in real time, and perform data fusion processing through the Kalman filter algorithm to improve data accuracy and reliability; b. Intelligent algorithm optimization module, which uses deep learning algorithms to extract features and diagnose faults from fused data, and uses reinforcement learning algorithms to optimize control strategies after a fault occurs, achieving fast and stable flight attitude adjustments; c. Environmental perception and adaptation module, which monitors the flight environment's meteorological parameters, topography, and electromagnetic interference in real time, and adjusts the parameters and thresholds of the fault diagnosis algorithm based on this environmental information to adapt to different environmental conditions; d. System integration and optimization module, used to deeply integrate the blade failure autonomous control system with the UAV's flight control system, power system, and communication system through hardware and software integration, achieving seamless docking and efficient communication.

2. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The multi-sensor fusion module includes: Acceleration sensor, used to monitor the vibration acceleration of the blade; Angular velocity sensor, used to monitor the rotational angular velocity of the blade; Strain sensor, used to monitor the strain of the blade; a vision sensor for capturing visual images of the blades; A data acquisition unit, used for synchronously collecting data from various sensors; The data processing unit uses the Kalman filter algorithm to perform data fusion processing.

3. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The intelligent algorithm optimization module includes a deep learning unit that uses a convolutional neural network (CNN) to extract features and diagnose faults from fused data, trains models using a large amount of labeled and unlabeled data, and a reinforcement learning unit that optimizes the drone's control strategy under different blade failure scenarios through interactive learning between the agent and the environment.

4. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The environment perception and adaptation module includes: Meteorological sensors to monitor temperature, humidity, air pressure, wind speed, and direction; Topographic mapping sensors, used to obtain topographic information; Electromagnetic interference sensor, used to detect electromagnetic field strength and interference source location; Adaptive control unit that adjusts the parameters and thresholds of the fault diagnosis algorithm according to environmental information.

5. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The system integration optimization module includes: Hardware interface unit, using standardized hardware interface; Communication protocol unit, using CAN bus or serial communication protocol; The software integration unit deeply integrates the fault diagnosis software with the flight control software and ground station software.

6. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The data acquisition frequency of the multi-sensor fusion module is not less than 100 Hz to ensure the real-time and accuracy of the data.

7. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The training data set of the deep learning algorithm includes at least 1,000 sets of normal operation data and 1,000 sets of fault operation data to improve the generalization ability of the model.

8. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The training environment of the reinforcement learning algorithm includes simulating flight scenarios under different wind speeds and load conditions to ensure the robustness of the control strategy.

9. The multi-rotor UAV blade failure autonomous control system according to claim 1, characterized in that: The communication delay of the system integration optimization module does not exceed 50ms to ensure the rapid execution of control instructions.

Images