eVTOL Flight Trainee Training System and Method Based on Virtual Low-Altitude Environment

Abstract

This invention discloses an eVTOL flight trainee training system based on a virtual low-altitude environment, relating to the field of flight training technology. It includes a data layer, a business logic layer, and a user interface layer. The data layer comprises a data source, a data synchronization center, a data transmission protocol, a data receiver, and a data storage module. The data source includes sensor data from flight trainee training equipment and low-altitude aircraft. The data synchronization center employs a distributed architecture to coordinate data synchronization across multiple components. This invention also discloses an eVTOL flight trainee training method based on a virtual low-altitude environment. The eVTOL flight trainee training system and method proposed in this invention, through innovative architecture design and functional implementation, effectively solve the problems of integration, scheduling, synchronization, and communication of various elements in a virtual environment. It provides an efficient and high-quality training platform for eVTOL flight trainees, contributing to the cultivation of flight personnel and promoting the application of aircraft in the transportation sector.

Description

Technical Field

[0001] This invention relates to the field of flight training technology, specifically to an eVTOL flight trainee training system and method based on a virtual low-altitude environment. Background Technology

[0002] With the acceleration of urbanization, eVTOL aircraft, as an important component of future urban air transportation, have increasingly broad application prospects. This emerging field has placed higher demands on the professional training of flight trainees. eVTOL aircraft not only need to operate flexibly in complex and ever-changing low-altitude environments, but also need to have the ability to fly in coordination with various low-altitude aircraft. Therefore, it is particularly urgent to develop a professional training system that can efficiently simulate real low-altitude flight environments and cover diverse flight scenarios. This system needs to have highly integrated elements, including accurate flight simulation, environment rendering, aircraft interaction, and real-time data feedback, in order to comprehensively improve the skill level of flight trainees and their ability to cope with complex situations.

[0003] However, existing flight training systems have significant shortcomings in simulating eVTOL flight. Traditional flight simulators struggle to accurately reproduce the complexity and variability of low-altitude environments, especially when dealing with elements such as buildings, terrain obstacles, and airflow changes. Furthermore, existing systems lack sufficient flexibility and synchronization in constructing virtual environments, resulting in monotonous flight simulation scenarios and low accuracy, failing to meet the needs of eVTOL flight trainees for high-fidelity and diverse training scenarios. At the same time, there are also obvious shortcomings in element integration, scheduling, synchronization, and communication, which limit the improvement of training effectiveness. In response, we propose an eVTOL flight trainee training system and method based on a virtual low-altitude environment. Summary of the Invention

[0004] To address the aforementioned technical issues, this technical solution provides an eVTOL flight trainee training system and method based on a virtual low-altitude environment, thus resolving the problems described above.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The eVTOL flight trainee training system based on a virtual low-altitude environment includes: a data layer, a business logic layer, and a user interface layer.

[0007] The data layer includes a data source, a data synchronization center, a data transmission protocol, a data receiver, and a data storage module. The data source includes sensor data from flight trainee training equipment and low-altitude aircraft. The data synchronization center adopts a distributed architecture to coordinate the data synchronization of multiple components. The data transmission protocol supports efficient encrypted transmission. The data receiver verifies and processes the data. The data storage module establishes a persistent storage mechanism.

[0008] The business logic layer implements functions such as virtual low-altitude scenario construction, aircraft simulation, student operation processing, digital pilot control, digital air traffic controller integration, and training evaluation. The virtual scenario construction is based on high-precision geographic information and meteorological models. The aircraft simulation includes physical characteristic modeling of eVTOL and other low-altitude aircraft. The student operation processing analyzes the control input in real time and feeds it back to the virtual environment. The digital pilot simulates collaborative flight behavior through AI algorithms. The digital air traffic controller generates instructions and coordinates traffic according to control rules.

[0009] The user interface layer provides an operating interface for instructors and an interaction interface for students. The system adopts a microservice architecture, which breaks down the functional modules into scenario management microservice, aircraft simulation microservice, student operation processing microservice, digital pilot service, digital air traffic controller service, evaluation feedback microservice and data processing microservice. Each microservice communicates through the DDS protocol.

[0010] Trainee training equipment connects to the system through a standardized interface, receiving virtual scenario data and uploading operational data; digital pilots and digital air traffic controllers are integrated into the system through independent interfaces, acquiring environmental data in real time and providing feedback on control commands.

[0011] Preferably, the data synchronization model adopts a synchronization method combining streaming and batch processing, specifically including the following steps:

[0012] Data Reading and Encapsulation: At each data source end of the training system, the data reading thread obtains the latest data from the data source in real time. After reading the data, it encapsulates it into standard data units. The data source reading threads read the data at time intervals Δ. Periodically collect sensor and operational data and encapsulate them into data units. ,in For timestamps, For sensor type identification, It is a numerical vector. For data verification codes, each Standardized data units corresponding to a single data collection;

[0013] Piping: Encapsulated data units are transmitted via a multi-threaded pipe. The read and write threads communicate through the pipe. Each time a read thread reads a result, it immediately sends it to the pipe without waiting for other data. The pipe capacity is... satisfy ,in For the number of concurrent data sources, For a single data source generated per unit of time Total number of data units The total data volume from all concurrent data sources is used to ensure that the pipeline capacity can handle all concurrent data, and the transmission latency meets the requirements. ;

[0014] Batch processing: The write thread waits to receive data at the other end of the pipe, and the write thread executes data according to a time window. Or data volume threshold Merge batches if the current cumulative data volume satisfy or Then these data will be merged into a single batch, generating a batch. ,in ;

[0015] Data transmission and processing: Once the data batch is ready, the write thread transmits the batch data to the target node via the network interface using an encryption protocol. Upon receiving the data batch, the target node parses and processes it, updates the system status after parsing, and verifies data consistency through a verification function. Verification, among which For hash functions, It is a cyclic redundancy check code.

[0016] Preferably, the data synchronization model combines distributed locks to achieve consistency control, specifically including:

[0017] Proposal generation: When sharing data When an update is needed, a component acts as a proposer, and the proposer generates a unique proposal number. And send a broadcast preparation request to most recipients. To the majority of recipients, among whom, A unique proposal number is generated using a hash function. Generate a timestamp for the proposal. A unique identifier for the proposer component;

[0018] Commitment Response: Recipients are distributed across key nodes in the system. Upon receiving a proposal's preparation request, they compare the proposal number with all previously responded-to proposal numbers. With the largest historical number ,like Then return the promise. Otherwise, refuse;

[0019] Proposal Confirmation: Recipients are located at key nodes in the system. Once the proposer receives confirmation from enough recipients, the accepted proposal is notified to all learners, who then select... or Send accept request ;

[0020] Status update: Recipient confirmed Persistence after not being overridden by higher proposals Learners record this proposal, learner nodes apply the update to their local state machine, and then propagate it to the entire system via a broadcast protocol to ensure eventual consistency is achieved. .

[0021] Preferably, the virtual low-altitude scene construction function utilizes geographic information data and graphics rendering technology to construct a virtual low-altitude environment, including diverse terrains such as cities, mountains and rivers, as well as high-rise buildings, bridges and low-altitude obstacles, to simulate meteorological conditions, including wind direction and speed, rain and snow, and fog, and dynamically adjusts meteorological parameters according to training needs;

[0022] It supports the rapid construction of various low-altitude mission scenarios, including urban aerial sightseeing, emergency medical rescue, and logistics delivery scenarios, each with specific mission objectives and environmental settings;

[0023] The virtual low-altitude scene construction function is based on a 3D geographic grid model. ,in For terrain elevation data, The set of obstacle coordinates; the meteorological simulation uses the Navier-Stokes equations to describe the fluid dynamics:

[0024] ;

[0025] in For wind speed vectors, For air pressure, The viscosity coefficient, External forces are used; dynamic meteorological parameters drive changes in the virtual environment through real-time equation solving, supporting dynamic adjustments to wind direction, rainfall intensity, and visibility.

[0026] Preferably, the aircraft simulation function accurately simulates the flight characteristics of the eVTOL aircraft, including the power response and handling performance in vertical takeoff and landing, hovering, and cruise flight modes, based on a real physical model and performance parameters;

[0027] The aircraft simulation function establishes a six-degree-of-freedom dynamic model for eVTOL:

[0028] ;

[0029] in Linear acceleration, Angular acceleration, For quality, For the inertial tensor, For propulsion, For aerodynamic drag, For gravity, For torque, Angular velocity; other aircraft simulations use simplified models. ,in, To accelerate the aircraft, For proportional gain, To preset the speed, This is the aircraft's current speed.

[0030] Preferably, the flight trainee operation processing function receives operation inputs from trainees on the training equipment, including flight attitude control, speed adjustment, and flight path planning, and converts them into aircraft actions in the virtual environment;

[0031] The student operation processing function uses a PID controller to process the control input. Transformed into aircraft control commands:

[0032] ;

[0033] in, To output control commands for adjusting aircraft maneuvers. For error signals, For integral gain, For differential gain, For the desired state, This is the current state; the system monitors operational compliance in real time, and will detect if the joystick input exceeds the safety threshold. If so, an alarm will be triggered and the deviation will be recorded. .

[0034] Preferably, in the digital pilot control function, the digital pilot autonomously pilots other low-altitude aircraft to fly in a virtual environment according to a preset flight plan and real-time environmental information;

[0035] Digital pilots make flight decisions based on air traffic rules and actual flight conditions, including avoiding other aircraft and choosing the optimal flight path;

[0036] Digital pilots employ deep reinforcement learning models Its policy network outputs actions :

[0037] ;

[0038] in, Output the action at the current moment. For the Critic network, For exploration rate, Gaussian noise; state Includes environmental data, aircraft status, and mission objectives, and a reward function. Weighted safety distance With range efficiency .

[0039] Preferably, the digital air traffic controller control functions include: generating control instructions based on virtual low-altitude traffic conditions and control rules; managing voice and data link communication using technologies such as speech synthesis; monitoring the situation in real time and optimizing control decisions; and rapidly coordinating resources from all parties to execute rescue plans in simulated emergency situations.

[0040] Digital air traffic controllers generate control instructions based on conflict detection algorithms and predict aircraft trajectory conflicts using the speed barrier method.

[0041] ;

[0042] In the formula, The closest approach point is used to predict the closest distance between two aircraft. , For the two aircraft in the future The position vector;

[0043] If the minimum proximity distance Then a height adjustment command is generated. Voice communication uses the WaveNet model to synthesize command audio. ,in As latent variables, Encoding for text.

[0044] Preferably, the training assessment function records various operational data of trainees during simulated flight, including flight trajectory, changes in flight attitude, and reaction time to emergencies;

[0045] By comparing the preset evaluation criteria with excellent flight operation cases, the trainees' performance is comprehensively evaluated. The evaluation indicators cover flight skills, decision-making ability, and safety awareness.

[0046] Generate a detailed evaluation report, including the trainees' strengths and weaknesses, and provide personalized improvement suggestions and follow-up training plans;

[0047] The training evaluation function uses a weighted scoring system based on multiple indicators:

[0048] ;

[0049] In the formula, For the accuracy of flight trajectory, For safety reasons, For flight efficiency, among which , , Weight The evaluation report generates decision tree classification results and marks the nodes with operational defects.

[0050] A method for eVTOL flight trainee training based on a virtual low-altitude environment, used to implement the eVTOL flight trainee training system based on a virtual low-altitude environment, includes:

[0051] Load geographic data into memory matrix Initialize the microservice instance and register it with the service registry, establish a secure channel via the TLS 1.3 protocol, and verify the device certificate. Digital pilots select actions according to Markov decision-making processes, and digital air traffic controllers perform actions every Δ... Update control orders and conduct simulation training;

[0052] The gradient boosting tree model is used to analyze operational data and generate personalized improvement paths. Complete the evaluation and feedback.

[0053] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0054] The eVTOL flight trainee training system and method based on a virtual low-altitude environment proposed in this invention effectively solves the problems of integration, scheduling, synchronization and communication of various elements in the virtual environment through innovative architecture design and functional implementation. It provides an efficient and high-quality training platform for eVTOL flight trainees, helps to cultivate high-quality eVTOL flight talents, and promotes the widespread application of eVTOL aircraft in the future transportation field. Attached Figure Description

[0055] Figure 1 Architecture diagram of a virtual low-altitude environment training system;

[0056] Figure 2 This is a data flow diagram of the present invention;

[0057] Figure 3 This is an example diagram of a simulated scenario of the present invention;

[0058] Figure 4 This is a flowchart illustrating the data synchronization process of the present invention. Detailed Implementation

[0059] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0060] Reference Figure 1 As shown, the eVTOL flight trainee training system based on a virtual low-altitude environment comprises a data layer, a business logic layer, and a user interface layer. These layers collaborate to achieve the functions of the flight trainee training system.

[0061] Data layer:

[0062] Data sources: Data sources from trainee training equipment (such as joysticks, pedals, and other operation data acquisition devices) and other low-altitude aircraft data sources (such as various sensors) are the source of data, providing raw data for the entire system.

[0063] Data Synchronization Center: As the core hub for data processing, its internal modules such as data verification, format conversion, and batch processing ensure the accuracy and validity of data. For example, the data verification module checks the integrity and accuracy of data, the data format conversion module unifies data from different formats, and the data batch processing module groups data for processing, improving efficiency.

[0064] Data transmission protocols: such as DDS protocols ensure secure and efficient data transmission within the system.

[0065] Data receiving end: mainly receives data from the data transmission protocol and interacts with the student training equipment to achieve bidirectional data transmission, such as sending virtual low-altitude scene data to the student training equipment and receiving student operation data at the same time.

[0066] Data storage: Used to store and retrieve data, ensuring data persistence and providing historical data support for the system.

[0067] Business logic layer:

[0068] Scene Management Microservice: Responsible for managing information related to virtual scenes, such as scene switching and scene parameter settings. It works closely with the Flight Simulation Microservice to provide scene information.

[0069] The aircraft simulation microservice includes an eVTOL simulation submodule and other low-altitude aircraft simulation submodules. It simulates the flight characteristics and behaviors of different types of aircraft, receives scenario information from the scenario management microservice and operation instructions from the student operation processing microservice, and feeds back flight data to the evaluation feedback microservice.

[0070] Trainee Operation Processing Microservice: Processes trainees' input on the training equipment, transmits the operation data to the aircraft simulation microservice, and sends the operation data to the evaluation feedback microservice for analysis.

[0071] Digital Pilot Service: Controls other low-altitude aircraft through a digital pilot (AI model), interacts with the aircraft simulation microservice to exchange flight control information, and feeds back flight status data to the evaluation feedback microservice.

[0072] Digital air traffic controller service: Generates control instructions based on information in the virtual low-altitude environment, sends them to the digital pilot service and aircraft simulation microservice, and receives aircraft status feedback and communication information to adjust control strategies in real time.

[0073] Evaluation and feedback microservice: Conducts comprehensive evaluation of student operation data, aircraft flight data, etc., and passes the evaluation data to the data processing microservice for storage and analysis.

[0074] Data processing microservices: responsible for processing and storing various types of data in the system, interacting with data storage in the data layer, and ensuring effective data management.

[0075] User interface layer:

[0076] Instructor interface: Instructors use this interface to manage and monitor the training system, such as setting training courses and adjusting training parameters.

[0077] Scene interaction interface: Provides trainees with an interface to interact with the virtual low-altitude environment, displaying information such as flight scenes and aircraft status.

[0078] eVTOL simulation submodule (student training equipment perspective): Students perform simulated flight operations on this equipment. The equipment interacts with the system through a standardized data interface, receiving virtual scene data and sending student operation data.

[0079] Reference Figure 2 As shown, the data flow includes:

[0080] 1. Data Acquisition and Transmission

[0081] Data sources include aircraft sensors (such as position and attitude sensors) and student operating equipment (control sticks, pedals), and the data they generate is the starting point of the entire process.

[0082] Data reading and encapsulation: Read data from the data source and encapsulate it, converting it into a format suitable for subsequent processing. The encapsulated data units are then sent to the pipeline transmission section.

[0083] Piping: A read thread continuously retrieves the latest data from the data source and sends it to the pipe, while a write thread receives the data at the other end of the pipe and processes it in batches. This mechanism ensures ordered data transmission and efficient processing.

[0084] Data batch processing: When certain conditions are met (such as time interval or data volume threshold), multiple data units are merged into a batch and then sent to the target node (such as the data processing module on the server side) through the network interface.

[0085] 2. Process based on distributed lock implementation

[0086] Proposer initiates proposal: When a batch of data arrives at the target node and involves shared data updates, the proposer (such as a digital air traffic controller or trainee training device) sends a preparation request to the majority of recipients, which includes the proposal number and the proposal content.

[0087] Recipient's processing of the preparation request: Upon receiving the request, the recipient compares the proposal number with the number of the responded proposal. If the received number is larger, the recipient accepts the proposal and promises not to accept proposals with smaller numbers. Then, the recipient replies with the following content regarding the processing of the preparation request:

[0088] The proposer sends an acceptance request: After receiving commitments from a majority of acceptors, the proposer sends an acceptance request. The proposal value in the request is usually the last proposal value accepted by the majority of acceptors. If it does not exist, a new proposal value is used.

[0089] Acceptance confirmation: After receiving the acceptance request, if the acceptor has not committed to the proposal with a higher number, the acceptor accepts the proposal and confirms the acceptance, and then notifies the learner of the accepted proposal.

[0090] Learner propagation of proposals: Learners record proposals and update the data therein to their local state machine to keep it consistent with the overall system state. They then propagate the updated data to other relevant components in the system to ensure data synchronization across the entire system.

[0091] Reference Figure 3 As shown, the simulated scenarios include:

[0092] 1. Virtual low-altitude environment

[0093] Terrain and Obstacles: Diverse terrains (cities, mountains, rivers) and obstacles (skyscrapers, bridges) create complex and realistic flight scenarios, providing trainees with a near-realistic training environment to hone their flight skills in dealing with different terrains and obstacles.

[0094] Aircraft Display: Various types of aircraft (eVTOL, helicopters, drones) are distributed in the virtual environment. Their flight direction and attitude changes demonstrate the simulated flight status. Numbers or names are used to distinguish and identify them, helping trainees understand how different aircraft operate in the same environment.

[0095] Meteorological conditions illustration: Complex meteorological conditions (wind direction and speed, rainfall, snowfall, fog) are presented through graphics and annotations. The dynamically adjustable meteorological parameters enable trainees to adapt to various weather conditions and improve their flight capabilities in complex meteorological environments.

[0096] 2. Demonstration of Digital Air Traffic Controllers and Digital Pilots

[0097] Digital air traffic controller console: Includes a display screen (showing the virtual low-altitude environment situation), communication equipment (voice and data link communication) and operation buttons (generating control instructions), which manages air traffic through information interaction and ensures safe and orderly flight.

[0098] Digital pilot-controlled aircraft: These aircraft are distinguished from those operated by trainees by special markings or colors. Their flight trajectories and attitude changes reflect preset flight strategies and algorithms, simulate different pilot behaviors, provide trainees with diverse training scenarios, and enhance their ability to fly collaboratively with other aircraft.

[0099] 3. Flight trainee perspective display (optional)

[0100] Trainee Training Equipment Perspective: Displays the virtual low-altitude scene seen by trainees on the training equipment, including aircraft instrument displays (speed, altitude, attitude, etc.) and partial views (terrain, obstacles, other aircraft, etc.), allowing trainees to intuitively experience the training content and enhancing their immersion in the training.

[0101] Reference Figure 4 As shown, the data synchronization process includes:

[0102] 1. Data Acquisition and Preprocessing Stage

[0103] Data sources include aircraft sensors (position, velocity, attitude sensors, etc.), student control equipment (joysticks and pedals), and meteorological and terrain data modules. These data sources provide various key data during flight, such as the aircraft's real-time status, student actions, weather conditions, and terrain information. They serve as the starting point for the entire data synchronization process, providing a rich and diverse data foundation for subsequent processing.

[0104] Data Cleaning and Labeling Module: Raw data obtained from the data source first enters this module. The data cleaning process removes noise and outliers from sensor data, and processes errors or invalid information in operational data to ensure data accuracy and reliability. The data labeling process classifies and labels the cleaned data, for example, labeling student operational data according to different operation types, so that subsequent models can better understand and process this data, preparing it for effective data utilization.

[0105] 2. Model Building and Training Phase

[0106] Hybrid Model Structure: The hybrid model based on RNN and LSTM is the core of the entire data synchronization process. The model's input receives pre-processed data containing information on aircraft status, student actions, environment, and other aspects, providing comprehensive data input. The RNN layer, through its unique neuron structure and recurrent connections, effectively extracts time-series features from the data and learns the dependencies between data over time. The LSTM layer, using mechanisms such as input gates, forget gates, and output gates, further filters and manages information, addressing long-term dependency issues and better capturing long-term patterns and trends in the data.

[0107] Training Process: The training set is used to train the model. Training data is input into the RNN layer to initiate the training process, and the model continuously adjusts its internal parameters to minimize prediction error. Internally, data flows from the RNN layer to the LSTM layer, enabling data transfer and processing between different layers. The final output provides the predicted synchronization data or data difference values ​​for subsequent data synchronization operations. The test set is used to evaluate the model's performance on unseen data. By comparing the predicted results at the output with the real data in the test set, various evaluation metrics (such as root mean square error and mean absolute error, mentioned later) are used to assess the model's accuracy and generalization ability, ensuring its reliability in real-world applications.

[0108] 3. Data synchronization and update phase

[0109] Real-time data input and processing: During system operation, data sources continuously generate real-time data (such as new data constantly collected by aircraft sensors during flight and real-time operation data from trainees). This data is directly input to the model's input side in the form of dashed arrows, participating in model processing along with preprocessed data. Based on this real-time data and the trained model parameters, the model predicts the synchronization adjustment amount or the synchronized data, thereby achieving synchronous processing of real-time data and ensuring that the data in the system always remains up-to-date and accurate.

[0110] Data Update and Distribution: The data update module receives the predicted synchronization data or data difference values ​​output by the model and applies this data to the actual system through its internal update calculation process. Specifically, the updated data is distributed to the virtual environment rendering module to update the display of the virtual flight scene in real time, allowing trainees to see a picture consistent with the actual flight status; it is also sent to the aircraft status update module to update the aircraft's status information in the simulation system, ensuring the accuracy of the simulated flight; and it is also transmitted to the digital air traffic controller information receiving module to provide air traffic controllers with the latest aircraft status and environmental information, enabling them to make accurate control decisions and achieve data synchronization and collaborative work among all components of the entire system.

[0111] 4. Model Evaluation and Adjustment Phase

[0112] Monitoring Points and Data Collection: Multiple monitoring points, such as Monitoring Point 1 and Monitoring Point 2, are set up at key locations in the system. These points are distributed along the data flow paths from the data source to the model input, from the model output to the data update module, and between various system components. These monitoring points are responsible for collecting various information related to model performance and data synchronization effectiveness, such as data transmission latency, model prediction accuracy, and data consistency. By monitoring this information in real time, potential problems in the system can be identified promptly, such as data transmission interruptions, excessive model prediction bias, and untimely data updates, providing a basis for subsequent adjustment decisions.

[0113] Adjustment Decision and Model Optimization: Information collected from monitoring points is aggregated into the adjustment decision module, which analyzes and evaluates this information through a logical judgment process. If a decline in model performance is detected (e.g., prediction error exceeds a preset threshold) or anomalies occur in data synchronization (e.g., excessively long data update delays or data inconsistencies), the adjustment decision module will make corresponding decisions based on preset rules. On one hand, it sends instructions to the model to fine-tune model parameters, optimizing the model's data processing capabilities and improving prediction accuracy by adjusting the internal parameters of the RNN and LSTM layers. On the other hand, it feeds new or adjusted training data back into the training set, enabling the model to learn the latest data patterns and trends, thereby continuously adapting to various situations during system operation. This achieves continuous model optimization and dynamic adjustment of the data synchronization process, ensuring the stable operation of the entire flight trainee training system and high-quality data synchronization.

[0114] The specific implementation method is as follows:

[0115] In terms of hardware configuration, if using a local server, choose a high-performance workstation or server host equipped with a multi-core high-performance CPU, large-capacity memory, high-speed solid-state drive, and professional graphics card to meet the needs of complex scene rendering and large-scale data processing. If using a cloud server, choose a cloud service instance with high computing power, large memory, and high-speed network bandwidth. Student training equipment can be a professional simulator cockpit equipped with a high-resolution display, precise flight joystick and pedals, or a high-performance PC with flight simulator peripherals. Network equipment should use high-speed Ethernet switches or enterprise-grade wireless routers to ensure stable and high-speed data transmission.

[0116] For software installation and configuration, a PostgreSQL database was installed on the server, and performance optimization and data structure design were performed. Relevant data tables were created to store aircraft data, scenario data, student operation records, etc. Various microservice programs were developed using suitable programming languages ​​(such as C++, Python, etc.) and development frameworks, deployed on the server, and the DDS communication environment was configured. Customized client software was installed on the student training equipment. This software can establish a stable connection with the server, receive and display virtual scenario data, collect student operation data, and transmit it to the server.

[0117] Detailed performance data for eVTOL and various low-altitude aircraft are collected, including airframe structural parameters, power system characteristics, and flight control system parameters. Modeling is then conducted using experimental testing and data provided by manufacturers. High-precision low-altitude geographic information data is acquired, such as terrain data obtained through satellite remote sensing and aerial mapping, and then digitally processed and optimized. Meteorological data is collected, and meteorological models are established to simulate weather changes under different climatic conditions. Air traffic rules and various flight mission procedures are compiled to provide a basis for digital pilots and digital air traffic controllers.

[0118] Instructors select appropriate training scenarios and tasks based on the training syllabus and the trainees' actual skill level. For example, they may choose simple flight missions in open areas for beginners, and gradually increase the complexity of the scenarios and the difficulty of the missions as the trainees' skills improve.

[0119] Once the trainees enter the training equipment and log into the system, the instructors begin the simulation training. During the training, the instructors can monitor the trainees' operations, the virtual scenario status, the digital pilot's behavior, and the digital air traffic controller's instructions in real time through the instructor's interface, adjusting scenario parameters, traffic flow, or providing guidance as needed.

[0120] Trainees conduct simulated flight operations according to flight operation procedures and mission requirements, experience flight challenges in different flight environments and mission scenarios, learn skills for coordinated flight with other low-altitude aircraft, and learn emergency response methods for dealing with emergencies (such as aircraft malfunctions, severe weather, air traffic conflicts, etc.), and follow the instructions of digital air traffic controllers.

[0121] After each simulated flight, the system automatically generates an evaluation report and transmits it to the trainee's training equipment and the instructor's terminal. Based on the evaluation results, the instructor analyzes the trainees' problems, such as unfamiliarity with flight operations, unclear understanding of mission objectives, and inadequate execution of air traffic control instructions, and adjusts the training plan accordingly, adding specialized exercises or theoretical explanations.

[0122] Regularly collect feedback from trainees and instructors regarding the system, and optimize and improve its performance, functionality, and scenarios. For example, based on trainees' feedback on the realism of the scenarios, improve the terrain and weather simulation effects; based on instructors' needs for training management, refine the instructor interface functions; and based on feedback on the accuracy of the digital air traffic controller simulation, optimize control rules and algorithms to improve training efficiency and quality.

[0123] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An eVTOL trainee pilot training system based on a virtual low-altitude environment, characterized by, Comprise: Data layer, business logic layer and user interface layer; The data layer includes data source, data synchronization center, data transmission protocol, data receiving end and data storage module, wherein the data source includes sensor data of flight student training equipment and low altitude aircraft, the data synchronization center adopts a distributed architecture to coordinate multi-component data synchronization, and corresponds to a data synchronization model, which specifically includes a combination of streaming and batch synchronization mode and consistency control based on distributed lock, the data transmission protocol supports efficient encryption transmission, the data receiving end verifies and processes data, and the data storage module establishes a persistent storage mechanism; The business logic layer realizes virtual low altitude scene construction, aircraft simulation, student operation processing, digital pilot control, digital air traffic controller integration and training evaluation functions, wherein the virtual scene construction is based on high-precision geographic information and weather model, the virtual low altitude scene construction function utilizes geographic information data and graphic rendering technology to construct a virtual low altitude environment, including diversified terrain such as cities, mountains and rivers, and low altitude obstacles such as high-rise buildings and bridges, simulates meteorological conditions including wind direction and speed, rainfall and snowfall and fog, and dynamically adjusts meteorological parameters according to training needs; Supports rapid construction of multiple low altitude task scenarios, including city air tour, emergency medical rescue and logistics distribution scenarios, each scenario has specific task objectives and environment settings; The virtual low-altitude scene construction function is based on a three-dimensional geographic grid model wherein is terrain elevation data, is an obstacle coordinate set; the weather simulation uses the Navier-Stokes equation to describe fluid dynamics: ； wherein is the wind velocity vector, is the air pressure, is the viscosity coefficient, is the external force; dynamic meteorological parameters drive the virtual environment changes by real-time solving equations, supporting dynamic adjustment of wind direction, rainfall intensity and visibility; aircraft simulation includes modeling of physical characteristics of eVTOL and other low-altitude aircraft, student operation processes real-time analysis of manipulation input and feedback to the virtual environment, digital pilots simulate collaborative flight behavior through AI algorithms, and digital air traffic controllers generate instructions and coordinate traffic according to control rules; The user interface layer provides instructor operation interface and student interaction interface; The system adopts micro-service architecture, and divides the functional modules into scene management micro-service, aircraft simulation micro-service, student operation processing micro-service, digital pilot service, digital air traffic controller service, evaluation feedback micro-service and data processing micro-service, and each micro-service communicates through DDS protocol; The student training equipment is connected with the system through a standardized interface, receives virtual scene data and uploads operation data; The digital pilot and the digital air traffic controller are integrated into the system through independent interfaces, and real-time environment data is obtained and control instructions are fed back.

2. The virtual low-altitude environment-based eVTOL trainee pilot training system and method of claim 1, wherein, The data synchronization model adopts a combination of streaming and batch synchronization mode, specifically including the following steps: Data reading and packaging: at each data source end of the training system, the data reading thread acquires the latest data from the data source in real time, and after reading the data, it is packaged into a standard data unit. The reading thread at the data source end reads the data at a time interval Δ Periodically collects sensor and operation data and packages it as a data unit wherein is a timestamp, is a sensor type identifier, is a numerical vector, is a data check code, and each corresponds to a single acquisition of standardized data unit; Pipeline transmission: the encapsulated data unit is transmitted through a multi-thread pipeline, the reading thread and the writing thread communicate through the pipeline, the reading thread sends the result data to the pipeline immediately after reading it, without waiting for other data, the pipeline capacity satisfies , wherein is the number of concurrent data sources, is the number of data units generated by a single data source per unit time, is the total amount of data units, is the total amount of data of all concurrent data sources, used to ensure that the pipeline capacity can carry all concurrent data, and the transmission delay satisfies ; Batch processing: The write thread at the other end of the pipe waits to receive data, the write thread merges batches if the current accumulated data amount or data volume threshold meets or or , the data is merged into a batch, and a batch is generated, where ; Data sending and processing: when the data batch is ready, the writing thread transmits the batch data to the target node through the network interface via an encryption protocol. After receiving the data batch, the target node parses and processes it. The target node updates the system state after parsing, and data consistency is checked by a verification function verification, wherein is a hash function, is a cyclic redundancy check code.

3. The virtual low-altitude environment-based eVTOL trainee pilot training system of claim 1, wherein, The data synchronization model realizes consistency control in combination with a distributed lock, specifically including: Proposal generation: when shared data needs to be updated, a certain component acts as a proposer, the proposer generates a unique proposal number , and sends a broadcast preparation request to the majority acceptors to the majority acceptors, wherein, is a unique proposal number, generated by a hash function, is a proposal generation timestamp, is a unique identifier of the proposer component; Promise response: acceptor distributed on the key node of the system, after receiving the preparation request of the proposer, compare the proposal number with all the proposal numbers that have been responded, acceptor compares with the largest number in history , if , return promise , otherwise reject; Proposal confirmation: the acceptors are distributed on the key nodes of the system, and when the proposer receives enough confirmations from the acceptors, the accepted proposal is notified to all learners, and the selection or , send an acceptance request ; State update: acceptors acknowledge After not being covered by a higher proposal, the persistent , the learner records this proposal, the learner node applies the update to the local state machine, and spreads it to the whole system through the broadcast protocol, ensuring that the eventual consistency is met .

4. The virtual low-altitude environment-based eVTOL trainee pilot training system of claim 1, wherein, The aircraft simulation function accurately simulates the flight characteristics of the eVTOL aircraft, including power response, control performance in vertical take-off and landing, hovering and cruise flight modes based on real physical models and performance parameters; The aircraft simulation function establishes a six-degree-of-freedom dynamics model for eVTOL: ； wherein is linear acceleration, is angular acceleration, is mass, is inertia tensor, is propulsion force, is aerodynamic drag, is gravity, is moment, is angular velocity; other aircraft models employ simplified models wherein, is aircraft acceleration, is proportional gain, is preset airspeed, is current aircraft airspeed.

5. The virtual low-altitude environment-based eVTOL student pilot training system of claim 1, wherein, The flight student operation processing function receives student operation input on the training equipment, including flight attitude control, speed adjustment and route planning, and converts it into aircraft action in the virtual environment; The student operation processing function converts the manipulation input into aircraft control commands: ； wherein, is an output control instruction for adjusting the aircraft action, is an error signal, is an integral gain, is a derivative gain, is a desired state, is a current state; the system monitors the operation compliance in real time, if the joystick input exceeds the safety threshold , an alarm is triggered and the deviation amount is recorded .

6. The virtual low-altitude environment-based eVTOL trainee pilot training system of claim 1, wherein, In the digital pilot control function, the digital pilot autonomously pilots other low altitude aircraft in the virtual environment according to the preset flight plan and real-time environment information; The digital pilot makes flight decisions according to air traffic rules and actual flight conditions, including avoiding other aircraft and selecting the optimal flight path. Digital pilot employing deep reinforcement learning model whose policy network outputs actions : ； wherein, is the action output at the current time, is the Critic network, is the exploration rate, is the Gaussian noise; state contains environment data, aircraft state, and mission objectives, reward function weighted safety distance with voyage efficiency .

7. The virtual low-altitude environment-based eVTOL trainee pilot training system of claim 1, wherein, The digital air traffic controller control function includes: generating control instructions based on virtual low-altitude traffic conditions and control rules; using voice synthesis technology for voice and data link communication management; monitoring the situation in real time and optimizing control decisions; coordinating resources to execute rescue plans when simulating emergency situations; The digital air traffic controller generates control instructions based on conflict detection algorithms and uses the speed barrier method to predict aircraft trajectory conflicts: ； wherein is the closest approach, used to predict the closest distance of two aircraft, , is the position vector of two aircraft at a future time . If the minimum approach distance , then generate height adjustment instructions ; voice communication uses WaveNet model to synthesize instruction audio , wherein is a latent variable, is a text encoding.

8. The virtual low-altitude environment-based eVTOL trainee pilot training system of claim 1, wherein, The training evaluation function records the student's operation data in the simulation flight, including flight trajectory, flight attitude change, and reaction time to unexpected situations; Compared with the preset evaluation standard and excellent flight operation case, the student's performance is comprehensively evaluated, and the evaluation index covers flight skill, decision-making ability, and safety awareness; Generate a detailed evaluation report, including the strengths and weaknesses of the student, and provide personalized improvement suggestions and follow-up training plans; The training evaluation function is weighted by multiple indicators: ； In the formula, For flight trajectory accuracy, For safety factor, For flight efficiency, wherein , , , weight ; the evaluation report generates a decision tree classification result, and labels the operation defect node.

9. A method of eVTOL trainee pilot training based on a virtual low altitude environment, characterized in that, For implementing the eVTOL flight student training system based on a virtual low-altitude environment as claimed in any one of claims 1-8, comprising: Loading geographic data into a memory matrix , initializing microservice instance and registering to service registry, establishing secure channel through TLS1.3 protocol, verifying device certificate ; digital pilot selects action according to Markov decision process, digital air traffic controller updates control instruction every Δ t; Analyzing operational data using a gradient boosting tree model to generate a personalized improvement path , completing the evaluation feedback.

Images