Aircraft control method and aircraft control device
Through the combination of deep learning and reinforcement learning algorithms, the aircraft's safe and stable flight and control strategies in complex environments are achieved, and the problem that traditional control methods are difficult to adapt to environmental changes is solved.
Patent Information
- Application Number
- CN202411975064.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-04-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
It is difficult for aircraft to achieve safe and stable flight in complex and changeable environments, and traditional control methods are difficult to adapt to environmental changes and emergencies.
Deep learning model combined with reinforcement learning algorithms is adopted to collect aircraft state and environmental data in real time, perform feature extraction and prediction, generate optimal control strategy, and dynamically adjust control instructions to optimize flight path and performance.
It realizes dynamic optimization of precise monitoring and control strategies for the aircraft state and environment, improves the safety and stability of the aircraft, and can respond in real time in emergencies.
Smart Images
Figure CN119828730A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of aircraft, and more specifically, to an aircraft control method and an aircraft control device. Background Art
[0002] With the rapid development of science and technology, aircraft, especially small aircraft such as drones, are increasingly used in civil and military fields. From environmental monitoring, logistics distribution to military reconnaissance and disaster relief, drones have become an important tool in modern mission execution with their flexibility and efficiency. However, with the continuous expansion of application scenarios, the control problem of aircraft has become increasingly prominent. How to ensure the safe and stable flight of aircraft in a complex and changing environment, while optimizing its flight path and performance parameters, has become a hot topic and difficulty in current research.
[0003] 1. Challenges of Aircraft Control
[0004] Environmental complexity: During flight, the aircraft needs to face complex and changing environmental conditions, such as changes in wind direction and speed, differences in terrain and landforms, and the sudden appearance of obstacles. These factors will affect the flight status of the aircraft and increase the difficulty of control.
[0005] Real-time status monitoring: In order to accurately grasp the flight status of the aircraft, it is necessary to collect various parameters of the aircraft in real time, such as altitude, speed, attitude angle, etc. At the same time, it is also necessary to obtain external environmental data in real time, such as obstacle distance, weather conditions, etc. The real-time and accuracy of these data are directly related to the safe flight and performance optimization of the aircraft.
[0006] Intelligence of control strategy: Facing complex environments and changing flight conditions, traditional control methods often fail to achieve ideal control effects. Therefore, it is necessary to introduce intelligent control algorithms, such as deep learning and reinforcement learning, to achieve accurate monitoring of the aircraft status and dynamic optimization of the control strategy.
[0007] Flight safety guarantee: During the flight, the aircraft may encounter various emergencies, such as engine failure, bad weather, etc. These emergencies may threaten the safety of the aircraft. Therefore, it is necessary to design an effective emergency avoidance mechanism to ensure the safety of the aircraft.
[0008] 2. Limitations of Existing Technologies
[0009] Limitations of sensor technology: Although existing sensor technology can realize real-time collection of aircraft status and external environment data, the accuracy, stability and reliability of sensors are still insufficient. Especially in complex environments, the performance of sensors is easily disturbed, resulting in inaccurate data.
[0010] Limitations of control algorithms: Traditional control algorithms are often based on fixed mathematical models and parameters, and are difficult to adapt to complex and changing environments and changing flight conditions. Although intelligent control algorithms have stronger adaptability and robustness, they have high computational complexity and real-time requirements, making them difficult to implement in practical applications.
[0011] Limitations of emergency hedging mechanisms: Existing emergency hedging mechanisms are often based on fixed thresholds and rules, making it difficult to achieve real-time response and effective handling of emergencies. At the same time, the design of emergency hedging mechanisms often lacks flexibility and is difficult to adapt to the needs of different scenarios. Summary of the invention
[0012] 1. Technical problems to be solved
[0013] In view of the problems existing in the prior art, the present invention provides an aircraft control method and an aircraft control device.
[0014] 2. Technical solution
[0015] To solve the above problems, the present invention adopts the following technical solutions.
[0016] A method for controlling an aircraft and an aircraft control device, comprising the following steps:
[0017] S1. Collect the status parameters and environmental data of the aircraft in real time, and use the deep learning model to extract features and classify the collected data;
[0018] S2. Based on the extracted features, the reinforcement learning algorithm is used to predict the future state of the aircraft and environmental changes, and generate the optimal control strategy;
[0019] S3. Based on the prediction results and the optimal control strategy, dynamically adjust the control instructions of the aircraft to optimize the flight path and performance.
[0020] As a further improvement of the present invention, the deep learning model adopts a combination of convolutional neural network (CNN) and recurrent neural network (RNN) to improve the accuracy of feature extraction and classification.
[0021] As a further improvement of the present invention, multiple layers of safety thresholds are set. When it is predicted that the aircraft state or environmental parameters exceed a certain layer of safety range, the corresponding emergency avoidance program is automatically triggered to ensure flight safety.
[0022] As a further improvement of the present invention, the multi-layer security threshold includes:
[0023] Warning thresholds, which are triggered when parameters approach but do not exceed the safety range, alerting the pilot or control system;
[0024] Danger threshold, triggered when the parameter exceeds the safety range but has not yet reached the emergency avoidance condition, the control system begins to prepare the emergency avoidance procedure;
[0025] The emergency avoidance threshold is triggered when the parameter seriously exceeds the safety range and the emergency avoidance procedure needs to be executed immediately.
[0026] An aircraft control device, comprising:
[0027] Control systems;
[0028] Data acquisition module, used to obtain the status parameters and environmental data of the aircraft in real time;
[0029] Deep learning processing unit, with built-in deep learning models for data processing, feature extraction and classification;
[0030] A reinforcement learning processing unit, which generates an optimal control strategy using a reinforcement learning algorithm based on the output of the deep learning processing unit;
[0031] The control instruction output module sends a control signal to the aircraft actuator according to the output of the reinforcement learning processing unit.
[0032] As a further improvement of the present invention, it also includes:
[0033] Wireless communication module, used for high-speed data transmission with ground stations or other aircraft, to achieve remote monitoring, collaborative flight and real-time information sharing.
[0034] As a further improvement of the present invention, the data acquisition module includes but is not limited to laser radar, millimeter wave radar, infrared sensor, and visual sensor to comprehensively and accurately monitor the flight status and external environment.
[0035] As a further improvement of the present invention, it also includes an emergency avoidance module, which includes a pair of avoidance tubes arranged along the flight direction of the aircraft, and the pair of avoidance tubes are symmetrically distributed about the aircraft, and are used to quickly extend out to protect the aircraft when the emergency avoidance is triggered.
[0036] As a further improvement of the present invention, the escape cylinder includes a main fixed cylinder, a matching secondary extension cylinder is slidably installed on the main fixed cylinder along the forward direction of the aircraft, a connected protective airbag membrane is fixedly installed on one end of the secondary extension cylinder away from the main fixed cylinder, a magnetic piston is slidably installed in the secondary extension cylinder, an electromagnet module is fixedly installed in the main fixed cylinder, a compression spring is fixedly connected between the magnetic piston and the inner wall of the secondary extension cylinder, a plurality of limiting grooves are provided at the outer end of the secondary extension cylinder, a plurality of electric push rod modules corresponding to the limiting grooves are installed inside the main fixed cylinder, and the electric push rod module is close to the outer area of the main fixed cylinder and extends into the limiting grooves.
[0037] As a further improvement of the present invention, a micro positioning module is installed in the secondary extension tube, and the micro positioning module, the electromagnet module and the electric push rod module are all connected to the control system.
[0038] 3. Beneficial effects
[0039] Compared with the prior art, the advantages of the present invention are:
[0040] (1) The present invention integrates a variety of high-precision sensors to collect aircraft status parameters and external environment data in real time, and uses a deep learning model to extract and classify features, thereby achieving comprehensive and accurate monitoring of the aircraft status and environment, providing strong support for the subsequent generation of control strategies.
[0041] (2) Based on the features extracted by the deep learning model, the present invention uses a reinforcement learning algorithm to predict the future state of the aircraft and environmental changes and generate an optimal control strategy. This can achieve accurate prediction of the aircraft state and dynamic optimization of the control strategy, thereby improving the safety and stability of the aircraft.
[0042] (3) The present invention sets multiple layers of safety thresholds, including warning thresholds, danger thresholds, and emergency avoidance thresholds. When the state parameters or environmental data of the aircraft exceed a certain safety range, the corresponding emergency avoidance program is automatically triggered, which can achieve real-time response and effective processing of emergencies and ensure the safety of the aircraft. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 This is a flow chart of the control method of the present invention;
[0044] Figure 2 It is a schematic diagram of the control device of the present invention;
[0045] Figure 3 It is a structural schematic diagram of the emergency avoidance module of the present invention;
[0046] Figure 4 is a cross-sectional view of the emergency avoidance module of the present invention;
[0047] Figure 5 It is a schematic diagram of the structure of the emergency avoidance module of the present invention before and after avoiding danger and falling off.
[0048] Description of the numbers in the figure:
[0049] 1. Hazard avoidance cylinder; 101. Main fixed cylinder; 102. Secondary extension cylinder; 103. Protective airbag membrane; 104. Electromagnet module; 105. Magnetic piston; 106. Compression spring; 107. Electric push rod module; 108. Limiting slide groove. DETAILED DESCRIPTION
[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments, and all other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making creative work are within the scope of protection of the present invention.
[0051] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "top / bottom" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are 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 position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.
[0052] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "mounted / connected", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0053] Embodiment 1:
[0054] See also Figure 1-2 , a method for controlling an aircraft, comprising the following steps:
[0055] S1: Real-time data collection and feature extraction
[0056] Data acquisition module: In this embodiment, the data acquisition module integrates multiple high-precision sensors such as lidar, millimeter-wave radar, infrared sensor, visual sensor, etc., which are used to obtain real-time status parameters of the aircraft such as altitude, speed, attitude angle, etc. and external environment data such as obstacle distance, wind direction and speed, weather conditions, etc.
[0057] Deep learning processing unit: This unit has a built-in deep learning model that combines the convolutional neural network (CNN) and the recurrent neural network (RNN). CNN is responsible for extracting spatial features from sensor data, such as the shape and texture of objects, while RNN is responsible for capturing time series features, such as the changing trend of wind speed and the trajectory of the aircraft. Through the combination of the two, the deep learning model can extract and classify data features more accurately.
[0058] S2: Future state prediction and optimal control strategy generation
[0059] Reinforcement learning processing unit: Based on the features extracted by the deep learning processing unit, the reinforcement learning processing unit uses reinforcement learning algorithms such as Q-learning, DQN Deep Q-Network or A3CAsynchronous Advantage Actor-Critic to predict the future state of the aircraft and environmental changes. These algorithms learn what control strategy can obtain the maximum cumulative reward, that is, the optimal control strategy, in different environments through continuous trial and error and reward mechanisms.
[0060] Prediction and optimization: The reinforcement learning algorithm predicts the possible state changes of the aircraft in the future based on the current state, historical data and environmental model, and generates a series of control strategies accordingly. Then, the algorithm selects the optimal control strategy by evaluating the expected effect of each strategy.
[0061] S3: Dynamic adjustment of control instructions
[0062] Control command output module: Based on the optimal control strategy generated by the reinforcement learning processing unit, the control command output module sends precise control signals to the aircraft's actuators such as engines and control surfaces to dynamically adjust the aircraft's flight path and performance parameters.
[0063] Real-time feedback and adjustment: During the flight, the data acquisition module continuously collects new data, and the deep learning processing unit and reinforcement learning processing unit continuously update the model and optimize the control strategy based on this data to ensure that the aircraft can adapt to environmental changes in real time.
[0064] Multi-layer safety threshold and emergency avoidance procedures
[0065] Warning threshold: When the aircraft's status parameters or environmental data are close to but do not exceed the set safety range, the warning mechanism is triggered and a reminder message is sent to the ground station or pilot through the wireless communication module.
[0066] Danger threshold: When the parameters exceed the safety range but have not yet reached the emergency avoidance conditions, the control system automatically enters the ready state and starts the initialization of the emergency avoidance program.
[0067] Emergency avoidance threshold: When the parameters seriously exceed the safety range, the emergency avoidance procedure is immediately triggered, such as automatic landing, return or other obstacle avoidance actions to ensure flight safety.
[0068] A control device for an aircraft, comprising:
[0069] 1. Control system
[0070] As the core of the entire device, the control system is responsible for coordinating the work of each module, ensuring the smooth transmission and processing of data, and the accurate execution of control instructions.
[0071] 2. Data acquisition module
[0072] The data acquisition module integrates a variety of high-precision sensors, including but not limited to:
[0073] LiDAR: used for high-precision terrain mapping and obstacle detection.
[0074] Millimeter-wave radar: provides long-range, high-resolution flight environment perception.
[0075] Infrared sensor: monitors the temperature distribution of objects around the aircraft to assist flight at night or in low visibility conditions.
[0076] Visual sensor: Combines high-definition cameras with image processing algorithms to achieve target recognition and tracking.
[0077] These sensors work together to comprehensively and accurately monitor the flight status and external environment, providing high-quality data input for the deep learning processing unit.
[0078] 3. Deep Learning Processing Unit
[0079] The deep learning processing unit has a built-in deep learning model that combines the convolutional neural network CNN and the recurrent neural network RNN. CNN is responsible for extracting spatial features from sensor data, such as the shape and texture of objects, while RNN is responsible for capturing time series features, such as changes in aircraft speed, fluctuations in wind direction and speed, etc. Through the combination of the two, the deep learning model can achieve accurate feature extraction and classification of collected data, providing strong support for the subsequent generation of control strategies.
[0080] 4. Reinforcement Learning Processing Unit
[0081] Based on the output of the deep learning processing unit, the reinforcement learning processing unit uses reinforcement learning algorithms such as Q-learning, DQN, A3C, etc. to predict the future state of the aircraft and environmental changes, and generate optimal control strategies. These strategies are designed to maximize the performance indicators of the aircraft such as flight speed, energy efficiency, etc., while ensuring flight safety.
[0082] 5. Control instruction output module
[0083] According to the optimal control strategy generated by the reinforcement learning processing unit, the control instruction output module sends precise control signals to the aircraft's actuators such as engines and control surfaces to dynamically adjust the aircraft's flight path and performance parameters.
[0084] 6. Wireless communication module
[0085] The wireless communication module uses advanced communication technologies, such as 5G and satellite communications, to achieve high-speed data transmission with ground stations or other aircraft. This not only helps to achieve remote monitoring and collaborative flight, but also ensures real-time information sharing and improves the collaborative efficiency and safety of flight missions.
[0086] To ensure flight safety, the aircraft intelligent control device of this embodiment also sets multiple safety thresholds, including:
[0087] Warning threshold: When the aircraft's status parameters or environmental data are close to but not beyond the safety range, the warning mechanism is triggered, sending a reminder message to the ground station or pilot through the wireless communication module, and adjusting the control strategy to prevent potential risks.
[0088] Danger threshold: When the parameters exceed the safety range but have not yet reached the emergency avoidance conditions, the control system automatically enters the ready state and starts the initialization of the emergency avoidance program, such as adjusting the flight altitude, speed, etc., to reduce potential threats.
[0089] Emergency avoidance threshold: When the parameters seriously exceed the safety range and the emergency avoidance procedure needs to be executed immediately, the control system will automatically take emergency measures, such as automatic landing, returning to a safe area, etc., to ensure the safety of the aircraft.
[0090] Embodiment 2:
[0091] See also Figure 3-5 For small aircraft such as drones, an emergency avoidance module can also be integrated. The emergency avoidance module includes a pair of avoidance cylinders 1 arranged along the flight direction of the aircraft, and the pair of avoidance cylinders 1 are symmetrically distributed about the aircraft, and are used to quickly extend to protect the aircraft when the emergency avoidance is triggered.
[0092] The installation position of the refuge tube 1 can be integrated on the body or on the bracket. The specific position is selected according to the structure of the drone, but it is guaranteed that the stability of the drone will not deteriorate after installation. All parts on the refuge tube 1 can be made of lightweight materials to reduce their own weight.
[0093] The escape cylinder 1 includes a main fixed cylinder 101, on which a matching secondary extension cylinder 102 is slidably installed along the forward direction of the aircraft, and a connected protective airbag membrane 103 is fixedly installed at one end of the secondary extension cylinder 102 away from the main fixed cylinder 101, a magnetic piston 105 is slidably installed in the secondary extension cylinder 102, an electromagnet module 104 is fixedly installed in the main fixed cylinder 101, a compression spring 106 is fixedly connected between the magnetic piston 105 and the inner wall of the secondary extension cylinder 102, a plurality of limiting grooves 108 are provided at the outer end of the secondary extension cylinder 102, and a plurality of electric push rod modules 107 corresponding to the limiting grooves 108 are installed inside the main fixed cylinder 101, and the electric push rod module 107 is close to the outer area of the main fixed cylinder 101 and extends into the limiting grooves 108.
[0094] A micro positioning module is installed in the secondary extension tube 102 , and the micro positioning module, the electromagnet module 104 , and the electric push rod module 107 are all connected to the control system.
[0095] When the drone suddenly encounters an obstacle or other unexpected situation that makes it impossible for it to avoid or has no time to avoid it, the electromagnet module 104 can respond instantly and be energized to apply a repulsive magnetic field, so that the magnetic piston 105 drives the secondary extension tube 102 to extend outward until it can no longer extend. At this time, the magnetic piston 105 continues to overcome the elastic force of the compression spring 106 and continues to move, and fills the gas in the secondary extension tube 102 into the protective airbag membrane 103 to cause it to expand. Before the drone encounters an obstacle, it can be pre-contacted to provide good protection and prevent the drone from directly colliding with the obstacle and causing damage. It is worth noting that when the danger threshold is reached, the electromagnet module 104 can be started in advance, but the magnetic field can be weakened to the point where it can push the secondary extension tube 102 to extend but is not enough to overcome the elastic force of the compression spring 106 for inflation. When the emergency avoidance threshold is reached, the magnetic field can be increased to trigger the inflation action, so that the reaction speed during emergency avoidance will be faster.
[0096] In addition, when the protective airbag membrane 103 is damaged or cannot be retracted normally, it will affect the flight of the drone. The autonomous shedding program is triggered after the image is collected by the visual sensor on the data acquisition module. It is worth noting that the shedding action can only be triggered after it is found that the protective airbag membrane 103 cannot be used normally and there is no object in the area below the drone after collecting data, thereby improving the safety during shedding. That is, the electric push rod module 107 is started to retract from the limiting slide groove 108. At this time, the secondary extension tube 102 will lose its limiting effect and can fall off from the main fixed tube 101 autonomously, and then rely on the micro positioning module for recovery.
[0097] The above is only a preferred specific implementation of the present invention; however, the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and its improved conception within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for controlling an aircraft, characterized in that: The following steps are involved: S1. Collect the status parameters and environmental data of the aircraft in real time, and use the deep learning model to extract features and classify the collected data; S2. Based on the extracted features, the reinforcement learning algorithm is used to predict the future state of the aircraft and environmental changes, and generate the optimal control strategy; S3. Based on the prediction results and the optimal control strategy, dynamically adjust the control instructions of the aircraft to optimize the flight path and performance.
2. The method for controlling an aircraft according to claim 1, characterized in that: The deep learning model uses a combination of convolutional neural network (CNN) and recurrent neural network (RNN) to improve the accuracy of feature extraction and classification.
3. A method for controlling an aircraft according to claim 1 or 2, characterized in that: Multiple layers of safety thresholds are set. When it is predicted that the aircraft status or environmental parameters exceed a certain safety range, the corresponding emergency avoidance procedure is automatically triggered to ensure flight safety.
4. The method for controlling an aircraft according to claim 3, characterized in that: Multi-layer security thresholds include: Warning thresholds, which are triggered when parameters approach but do not exceed the safety range, alerting the pilot or control system; Danger threshold, triggered when the parameter exceeds the safety range but has not yet reached the emergency avoidance condition, the control system begins to prepare the emergency avoidance procedure; The emergency avoidance threshold is triggered when the parameter seriously exceeds the safety range and the emergency avoidance procedure needs to be executed immediately.
5. A control device for an aircraft, characterized in that: include: Control systems; Data acquisition module, used to obtain the status parameters and environmental data of the aircraft in real time; A deep learning processing unit, having a built-in deep learning model as claimed in claim 2, for data processing, feature extraction and classification; A reinforcement learning processing unit, which generates an optimal control strategy using a reinforcement learning algorithm based on the output of the deep learning processing unit; The control instruction output module sends a control signal to the aircraft actuator according to the output of the reinforcement learning processing unit.
6. The control device for an aircraft according to claim 5, characterized in that: Also includes: Wireless communication module, used for high-speed data transmission with ground stations or other aircraft, to achieve remote monitoring, collaborative flight and real-time information sharing.
7. The control device for an aircraft according to claim 5, characterized in that: The data acquisition module includes but is not limited to laser radar, millimeter wave radar, infrared sensor, and visual sensor to comprehensively and accurately monitor the flight status and external environment.
8. The control device for an aircraft according to claim 5, characterized in that: It also comprises an emergency avoidance module, which comprises a pair of avoidance cylinders (1) arranged along the flight direction of the aircraft, and the pair of avoidance cylinders (1) are symmetrically distributed about the aircraft, and are used to quickly extend out to protect the aircraft when the emergency avoidance is triggered.
9. The control device of an aircraft according to claim 8, characterized in that: The safety cylinder (1) comprises a main fixed cylinder (101), a matching secondary extension cylinder (102) is slidably mounted on the main fixed cylinder (101) along the forward direction of the aircraft, a connected protective airbag membrane (103) is fixedly mounted on one end of the secondary extension cylinder (102) away from the main fixed cylinder (101), a magnetic piston (105) is slidably mounted in the secondary extension cylinder (102), and an electromagnet module (104) is fixedly mounted in the main fixed cylinder (101). ), a compression spring (106) is fixedly connected between the magnetic piston (105) and the inner wall of the secondary extension tube (102), a plurality of limiting slide grooves (108) are opened at the outer end of the secondary extension tube (102), and a plurality of electric push rod modules (107) corresponding to the limiting slide grooves (108) are installed inside the main fixed tube (101), and the electric push rod module (107) is close to the outer area of the main fixed tube (101) and extends into the limiting slide groove (108).
10. The control device of an aircraft according to claim 9, characterized in that: A micro positioning module is installed in the secondary extension tube (102), and the micro positioning module, the electromagnet module (104) and the electric push rod module (107) are all connected to the control system.