Helicopter pilot driving skill intelligent aided training method and system

Abstract

The invention relates to the technical field of flight training, and particularly discloses an intelligent auxiliary training method and system for the driving skill of a helicopter pilot, and the method comprises the steps: deploying a plurality of sensors in a simulated helicopter cockpit for collecting operation, physiology, eye movement and other original data; after data standardization preprocessing, constructing a trainee training state comprehensive data model by using a multi-modal data fusion algorithm; dynamically generating a simulated flight scene by using VR and AR; carrying out comprehensive evaluation by adopting a multi-modal evaluation model based on deep learning in combination with multi-source data and student scene performance, and generating a multi-dimensional report; an evaluation result is fed back in real time through multiple channels such as voice and touch, and customized training resources are pushed; and formulating an exclusive training plan by using a personalized learning path planning algorithm based on multiple training data. Through multi-source data fusion and an intelligent algorithm, the training state of the trainee is comprehensively and accurately evaluated, a personalized training scheme is provided, and the driving skill training efficiency and quality of the helicopter pilot are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of flight training, and in particular to an intelligent auxiliary training method and system for helicopter pilots' driving skills. Background Art

[0002] In the aerospace industry, helicopter pilot skill training is crucial, with its effectiveness directly impacting flight safety and mission execution. Traditional helicopter pilot skill training relies primarily on on-site instruction from instructors and fixed-mode simulation training equipment. During actual training, instructors must observe trainees' operations in real time and provide guidance based on their own experience. This approach not only requires a high number of instructors but also creates a highly subjective process, making it difficult to ensure uniform training standards and resulting in varying training quality for different trainees.

[0003] Although existing simulation training equipment can simulate some flight scenarios, it has many limitations. On the one hand, it lacks the comprehensive collection and analysis of students' operational data, physiological data, and eye movement data, and cannot comprehensively and accurately evaluate the students' actual status during the training process. It is difficult to gain an in-depth understanding of the students' behavioral habits, psychological changes, and attention distribution during the operation process, and thus it is impossible to formulate targeted training strategies. On the other hand, the traditional training model cannot provide personalized training programs based on the individual differences of students. All students use the same training content and process, which makes the training inefficient and makes it difficult for students to quickly improve their driving skills to meet actual flight needs. In addition, the generation of traditional training scenarios lacks flexibility and cannot be dynamically adjusted according to the students' real-time training performance and status. It is difficult to effectively train students' coping capabilities in complex and changing environments. Students may not be able to properly handle emergencies encountered during actual flights.

[0004] Therefore, an intelligent auxiliary training method and system for helicopter pilot driving skills has become an urgent problem to be solved. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an intelligent assisted training method and system for helicopter pilot driving skills. By integrating multi-source data and applying advanced deep learning and intelligent algorithms, it can achieve comprehensive perception and accurate evaluation of the trainees' training status, dynamically generate personalized simulated flight scenarios, and provide multi-channel real-time feedback and customized training plans, thereby significantly improving the efficiency and quality of helicopter pilot driving skills training, reducing training costs and risks, and cultivating high-quality helicopter pilots who are better able to adapt to actual flight mission requirements.

[0006] To solve the above technical problems, the present invention provides a technical solution: a method for intelligent auxiliary training of helicopter pilot driving skills, comprising the following steps:

[0007] S1. Deploy several different types of sensors in the simulated helicopter cockpit to collect raw data, and transmit the data to the data processing module in real time through the wireless transmission module;

[0008] S2. After receiving the raw data, the data processing module performs standardized preprocessing on each type of data. It then uses a multimodal data fusion algorithm to deeply fuse the preprocessed operational data, physiological data, and eye movement data to construct a comprehensive data model of the trainee's training status, and obtain the trainee's training status based on the comprehensive data model.

[0009] S3. Based on the obtained student training status, use VR and AR to dynamically generate simulated flight scenarios;

[0010] S4. Use a multimodal evaluation model based on deep learning to conduct a comprehensive evaluation based on the fused multi-source data and the students’ performance in dynamic simulation scenarios, and generate a detailed multi-dimensional evaluation report.

[0011] S5. Based on the comprehensive assessment results, multi-channel real-time feedback is provided through voice interaction systems, tactile feedback devices, display screens, and virtual reality interaction interfaces. Customized training courses, case studies, and simulation exercises are delivered based on the trainees' training progress and assessment results.

[0012] S6. Based on the multi-dimensional evaluation data and feedback information of students' multiple training sessions, a personalized learning path planning algorithm is used to develop an exclusive training plan for each student.

[0013] Furthermore, the sensors include a displacement sensor provided on the operating lever, a pressure sensor provided on the foot pedal, a posture sensor provided on the seat, an eye tracking sensor, an electroencephalogram sensor, a heart rate sensor, and a skin charge sensor.

[0014] Furthermore, the electroencephalogram sensor collects electrical activity signals from the student's brain to analyze the student's cognitive load and concentration level; the heart rate sensor monitors the student's heart rate changes in real time to reflect the student's emotional state and physical stress level; the skin conduction sensor detects the conductivity of the student's skin to assist in judging the student's emotional fluctuations.

[0015] Furthermore, in step S2, based on the comprehensive data model, the deviation value of the operation index is calculated, and the correlation between the trainee's physiological and psychological state and the operation performance is analyzed.

[0016] Furthermore, in step S3, if the trainee performs poorly in a specific operation, hovering training scenes under different wind speeds and terrain conditions are generated; if the trainee is under high stress, a relatively simple and relaxing flight scene is generated to relieve stress, and prompt elements are set in the scene to gradually guide the trainee to correct operation.

[0017] Furthermore, in step S4, the evaluation results of the comprehensive evaluation include evaluations of operational skills, adaptability, and psychological stability.

[0018] Furthermore, in step S5, the voice interaction system provides real-time conversational guidance based on the evaluation results; the tactile feedback device simulates abnormal conditions during helicopter flight through vibrations of different intensities and frequencies to remind trainees; the display screen and virtual reality interaction interface display operational deviations and improvement suggestions in the form of three-dimensional animations and highlighted labels.

[0019] Furthermore, the exclusive training plan dynamically adjusts the difficulty of training scenarios, the focus of training content, and the training schedule; regularly reviews and analyzes the training effects of students, continuously optimizes the training plan, and synchronizes relevant information to the coaching terminal.

[0020] The present invention also provides an intelligent auxiliary training system for helicopter pilot driving skills, which is used to implement the above method, including:

[0021] Multi-source data acquisition module: including operational sensors, physiological sensors, and eye tracking sensors, used to collect various data of trainees during training in real time;

[0022] Data fusion and processing module: Receives data from the multi-source data acquisition module, performs pre-processing and deep fusion, and performs analysis and calculation to build a comprehensive data model of the trainee's training status;

[0023] Dynamic scene generation module: Based on the results of the data fusion and processing module, it uses VR and AR to dynamically generate diverse and personalized simulated flight scenes;

[0024] Intelligent multimodal assessment module: uses a deep learning multimodal assessment model to comprehensively evaluate students' performance in dynamic scenarios, outputting multi-dimensional assessment results and detailed reports;

[0025] Real-time interactive feedback module: used to interact with trainees, provide guidance and push training resources;

[0026] Personalized training optimization module: Develop and adjust personalized training plans based on trainee training data, continuously optimize the training process, and interact with the coach's terminal for data;

[0027] Coach terminal: used for coaches to view students' training data, evaluation reports and training plans in real time, provide remote guidance, intervene in the training process, communicate with students, and provide personalized training suggestions.

[0028] The advantages of the present invention compared with the prior art are:

[0029] This invention achieves a comprehensive and accurate assessment of the trainee's training status by fusing operational data, physiological data, and eye movement data and combining it with a deep learning algorithm, thus overcoming the limitations of traditional methods.

[0030] The present invention formulates exclusive training plans according to individual differences of students, dynamically adjusts training content and difficulty, meets the training needs of different students, and improves the pertinence and efficiency of training.

[0031] The present invention utilizes VR and AR technologies to dynamically generate simulated flight scenarios, combined with multi-channel real-time feedback, to provide trainees with an immersive and interactive training experience, thereby enhancing their adaptability and operational skills to complex flight environments.

[0032] The present invention reduces ineffective training time. Through intelligent evaluation and real-time feedback, trainees can promptly identify and correct problems, thereby accelerating the skill improvement process.

[0033] The present invention reduces dependence on a large number of training resources, replaces part of actual flight training with simulated training, reduces training costs and risks, and improves the utilization rate of training resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The present invention is a flowchart of an intelligent auxiliary training method for helicopter pilot driving skills.

[0035] Figure 2 The present invention is a system block diagram of an intelligent auxiliary training system for helicopter pilot driving skills. DETAILED DESCRIPTION

[0036] Various exemplary embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be noted that unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention.

[0037] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.

[0038] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0039] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0040] The following is a further detailed description of the intelligent auxiliary training method and system for helicopter pilot driving skills of the present invention in conjunction with the accompanying drawings.

[0041] Combined with attachment Figure 1-2 , the present invention is introduced in detail.

[0042] A method for intelligently assisting helicopter pilots in driving skills training comprises the following steps:

[0043] S1. Data collection steps: Various types of sensors are deployed in the simulated helicopter cockpit. Operational sensors include displacement sensors on the joystick, pressure sensors on the pedals, and posture sensors on the seat, which are used to collect motion data when the trainee operates the helicopter. Physiological sensors include electroencephalogram (EEG) sensors, heart rate sensors, and electrodermal sensors, which are used to collect brain electrical activity signals, heart rate changes, and skin conductivity, respectively, to analyze the trainee's cognitive load, emotional state, and physical stress level. Eye tracking sensors are used to collect eye movement data and obtain information about the trainee's gaze movement. These sensors transmit the collected raw data to the data processing module in real time via a wireless transmission module according to a specific data protocol.

[0044] S2. Data processing and fusion steps:

[0045] (1) After receiving the raw data, the data processing module uses a standardized method to preprocess each type of data to eliminate data dimension differences and noise interference, as follows:

[0046] in, is the standardized data, x i is the original data, μ x and σ x are the mean and standard deviation of the corresponding data series, respectively.

[0047] (2) Using a multimodal data fusion algorithm and a fusion model based on an attention mechanism, we can deeply fuse the operation data, physiological data, and eye movement data. First, we use a multi-layer perceptron (MLP) to extract the features of each modal data and obtain:

[0048]

[0049] Among them, MLP represents multi-layer perceptron, h o 、h p 、h e are the feature representations of each mode respectively.

[0050] (3) Calculate the attention weight of each modal data:

[0051]

[0052] in, W is the matrix weight and b is the bias parameter.

[0053] (4) The fusion feature is expressed as:

[0054] h fusion =α o h o +α p h p +α e h e ;

[0055] (5) Based on this, a comprehensive data model of the trainee’s training status is constructed, and the trainee’s training status is obtained through the model:

[0056] S=Softmax(MLP fusion (h fusion ));

[0057] Where S = [s1, s2, s3, s4, s5] T It represents the state vector of the trainees in the five dimensions of operational skills, adaptability, psychological stability, attention and fatigue.

[0058] (6) Based on the comprehensive data model, calculate the deviation value of the operation indicator. By comparing the actual operation data of the trainees with the standard operation data, the deviation is quantified using methods such as mean square error:

[0059] ΔO=O standard -O actual ;

[0060] Among them, ΔO is the operating index deviation value O standard is the standard operating data, O actual Provide students with practical operation data.

[0061] (7) Analyze the relationship between students’ physiological and psychological states and operational performance, for example, by calculating correlation coefficients to explore the relationship between changes in physiological indicators and operational errors.

[0062] S3. Scenario Generation Step: Based on the student's training status, simulated flight scenarios are dynamically generated using virtual reality (VR) and augmented reality (AR) technologies. If a student performs poorly on a specific operation, such as significant deviation in hovering, hovering training scenarios with varying wind speeds and terrain conditions are generated. If a student is under high stress (based on physiological data, such as heart rate exceeding a certain threshold and abnormally elevated skin conductance), a relatively simple and relaxing flight scenario is generated to relieve stress. At the same time, prompt elements such as virtual guide lines and prompt boxes for operation steps are set in the scenario to gradually guide the student to correct the operation.

[0063] S4. Comprehensive evaluation step: A multimodal evaluation model based on deep learning is used to conduct a comprehensive evaluation by combining the fused multi-source data and the trainee performance in the dynamic simulation scenario. fusion Combined with the dynamic simulation scene coding vector C, the input vector X = [h fusion ; C], where the dynamic simulation scene encoding vector C is encoded by the terrain, weather, mission objectives and other elements in the scene through a single-hot encoding, and then passes through the multi-layer perceptron MLP C Dimensionality reduction is obtained, that is, C = MLP C (One-Hot(Scene));

[0064] Input the input vector X into a deep learning network consisting of multiple convolutional layers Conv, recurrent layers RNN, and fully connected layers FC:

[0065]

[0066] Among them, Z1 and Z2 are the outputs of the convolutional layer, H is the output of the recurrent layer, and Y is the output of the fully connected layer;

[0067] The model is trained using the loss function, which uses the weighted cross entropy loss function:

[0068]

[0069] Among them, N is the number of samples, M is the number of evaluation dimensions (such as operational skills, adaptability, psychological stability, etc.), w ij is the weight of the jth dimension of the i-th sample, y ij is the true label, is the model prediction value;

[0070] In the evaluation phase, the input vector X is input into the trained model to obtain the evaluation result vector R = [r1, r2, r3] T , r1, r2, and r3 correspond to the evaluation scores of operational skills, adaptability, and psychological stability, respectively, and generate a detailed multi-dimensional evaluation report.

[0071] S5. Real-time feedback and resource push: Based on the comprehensive assessment results, multi-channel real-time feedback is provided through the voice interaction system, tactile feedback device, display screen, and virtual reality interactive interface. The voice interaction system provides real-time conversational guidance based on the assessment results, using natural language processing technology to generate targeted voice guidance content. The tactile feedback device simulates abnormal helicopter flight conditions through vibrations of varying intensities and frequencies to alert trainees, with the vibration pattern related to the type and severity of the abnormal condition. The display screen and virtual reality interactive interface display operational deviations and improvement suggestions using 3D animations and highlighted icons. Simultaneously, based on the trainee's training progress and assessment results, customized training courses, case studies, and simulation exercises are pushed from the training resource database, with the course content and difficulty matching the trainee's assessment results.

[0072] S6. Training Plan Development Steps: Based on the multi-dimensional evaluation data and feedback from multiple training sessions, a personalized learning path planning algorithm is used to develop a customized training plan for each student. This customized training plan dynamically adjusts the difficulty of training scenarios, the focus of training content, and the training schedule. Regular review and analysis of student training results are conducted to continuously optimize the training plan, and relevant information is synchronized to the coach's terminal.

[0073] To implement the above method, the present invention also proposes an intelligent auxiliary training system for helicopter pilot driving skills, comprising:

[0074] Multi-source data acquisition module: Contains operational sensors, physiological sensors, and eye-tracking sensors, which are installed in the simulated helicopter cockpit according to a specific layout. It is used to collect various types of data from trainees during training in real time and send the data to the data fusion and processing module through the wireless transmission module.

[0075] Data fusion and processing module: Receives data from the multi-source data acquisition module, performs standardized preprocessing, multimodal data fusion and related analysis and calculation operations, builds a comprehensive data model of the trainees' training status, and provides data support for subsequent modules.

[0076] Dynamic scene generation module: Based on the results of the data fusion and processing module, using VR and AR technologies, according to the set scene generation rules, it dynamically generates diversified and personalized simulated flight scenes and transmits the scene information to the display device.

[0077] Intelligent multimodal evaluation module: uses a deep learning multimodal evaluation model to comprehensively evaluate students' performance in dynamic scenarios, output multi-dimensional evaluation results and detailed reports, and pass the results to the real-time interactive feedback module and personalized training optimization module.

[0078] Real-time interactive feedback module: interacts with students, provides guidance and pushes training resources through various hardware devices based on the evaluation results, and realizes real-time interaction with students.

[0079] Personalized training optimization module: Based on the trainee training data, use the personalized learning path planning algorithm to formulate and adjust the personalized training plan, continuously optimize the training process, and interact with the coach terminal for data to receive coach instructions and feedback information.

[0080] Coach terminal: Coaches can use this terminal to view students' training data, evaluation reports and training plans in real time, provide remote guidance, intervene in the training process, communicate with students, provide personalized training suggestions, and realize remote management and supervision of students' training.

[0081] The specific implementation process of the intelligent auxiliary training method and system for helicopter pilot driving skills of the present invention is as follows:

[0082] 1. Multi-source data collection

[0083] In the simulated helicopter cockpit, the joystick displacement sensor uses a high-precision Hall-type displacement sensor, which is installed next to the joystick's rotation axis. It can accurately measure the displacement changes of the joystick in three-dimensional space with an accuracy of 0.01 mm; the pedal pressure sensor uses a piezoresistive pressure sensor, which is evenly distributed on the pedal surface and can detect pressure changes in real time. The measurement range is 0-2000N and the accuracy is 1N; the seat posture sensor uses an inertial measurement unit composed of a three-axis accelerometer and a gyroscope, fixed at the center of the bottom of the seat, which can measure the acceleration and angular velocity of the seat in three axes, with a sampling frequency of 100Hz.

[0084] The EEG sensor uses dry-electrode EEG acquisition equipment, and electrodes are fixed at multiple electrode sites, including Fp1, Fp2, F3, and F4, according to the international 10-20 system electrode placement standard, with a sampling frequency of 256Hz. The heart rate sensor uses a wrist-mounted photoelectric heart rate sensor, which monitors heart rate in real time using the PPG principle, with a sampling frequency of 1Hz. The galvanic skin sensor uses electrodes attached to the trainee's index and middle fingers to collect skin conductivity, with a sampling frequency of 1Hz. The eye tracking sensor uses an infrared eye tracker, installed directly in front of the cockpit, approximately 50cm from the trainee's eyes, and can track gaze movements in real time, with a sampling frequency of 60Hz. All sensors communicate with the data processing module via a Bluetooth 5.0 wireless module, and data transmission uses the TCP / IP protocol to ensure stable data transmission.

[0085] 2. Data fusion and processing

[0086] The data fusion and processing module was deployed on a high-performance server equipped with an Intel Xeon Gold 6348 processor, 128GB of RAM, a 4TB solid-state drive, and the CentOS 8 operating system. Data preprocessing was performed using Python scripts, utilizing the NumPy and SciPy libraries for data normalization, filtering, and other operations. Multimodal data fusion employed a deep learning model built on the TensorFlow framework. Model training was performed on a GPU cluster using the Adam optimizer with a learning rate of 0.001, a training batch size of 32, and 50 epochs. Model parameters were continuously adjusted to optimize data fusion performance and the accuracy of training status evaluation.

[0087] 3. Dynamic scene generation

[0088] The dynamic scene generation module is developed based on the Unity3D game engine, using C# to write scene generation logic scripts. The ProBuilder plug-in is used to generate diverse terrains, and the Weather Maker plug-in is used to simulate different weather conditions. When generating scenes, scripts dynamically adjust scene parameters such as terrain complexity, weather conditions, and obstacle distribution based on the trainee's training status data. Scene rendering utilizes the real-time ray tracing technology of NVIDIA RTX graphics cards to provide trainees with a realistic visual experience. After the scene is generated, it is presented to trainees through VR headsets (such as the HTC Vive Pro 2) and AR devices (such as the Microsoft HoloLens 2).

[0089] 4. Intelligent multimodal evaluation

[0090] The intelligent multimodal evaluation module builds a deep learning evaluation model based on the PyTorch framework. Model training data is sourced from a large number of student training records. After data cleaning and labeling, it is used for model training. During training, cross-validation is used to optimize model hyperparameters, and GPUs are used to accelerate the training process. Model evaluation metrics include accuracy, recall, and F1 score. Evaluation accuracy is improved by continuously optimizing the model structure and parameters. In actual application, the model receives data from the data fusion and processing module and the dynamic scene generation module in real time, rapidly outputting evaluation results.

[0091] 5. Real-time interactive feedback

[0092] The voice interaction system utilizes iFlytek's speech recognition and speech synthesis technologies, combined with the natural language processing framework NLTK, to deliver conversational voice guidance. The haptic feedback system uses an Arduino development board to control a vibration motor, adjusting the motor's vibration intensity and frequency based on the feedback. A 4K resolution LCD screen is installed in the front of the cockpit, and the virtual reality interface is implemented using a VR headset. Feedback content is generated using a Python script. Based on the evaluation results, appropriate content is selected from a library of pre-set feedback templates, personalized, and presented to the trainees.

[0093] 6. Personalized training optimization

[0094] The personalized training optimization module uses algorithms written in Python, with backend services developed using the Django framework. Student training data is stored in a MySQL database. Data mining algorithms (such as cluster analysis and association rule mining) are used to analyze this data and identify training patterns and weaknesses. The personalized learning path planning algorithm dynamically adjusts the training plan based on these analysis results, combined with training objectives and student characteristics. Adjusted training plans are synchronized to the coach's terminal and the student's training interface via a network interface, allowing students and coaches to view the updated training plan in real time.

[0095] 7. Coach Terminal

[0096] The coach terminal is developed as a web application, using the Vue.js framework for the front end and Spring Boot for the back end. Coaches log in to the system through a browser. The system provides an intuitive user interface that displays student training data, evaluation reports, and training plans in charts and tables. Coaches can communicate with students through the terminal by sending text and voice messages, using WebSocket technology for real-time communication. Coaches' interventions in training are transmitted via an API to the personalized training optimization module and the dynamic scenario generation module, enabling remote control and guidance of student training.

[0097] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A helicopter pilot driving skills intelligent auxiliary training method, characterized in that: The following steps are involved: S1. Deploy several different types of sensors in the simulated helicopter cockpit to collect raw data, and transmit the data to the data processing module in real time through the wireless transmission module; S2. After receiving the raw data, the data processing module performs standardized preprocessing on various types of data, applies a multimodal data fusion algorithm to deeply fuse the preprocessed data, constructs a comprehensive data model of the trainee's training status, and obtains the trainee's training status based on the comprehensive data model; S3. Based on the obtained student training status, use VR and AR to dynamically generate simulated flight scenarios; S4. Use a multimodal evaluation model based on deep learning to conduct a comprehensive evaluation based on the fused multi-source data and the students’ performance in dynamic simulation scenarios, and generate a detailed multi-dimensional evaluation report. S5. Based on the comprehensive assessment results, multi-channel real-time feedback is provided through voice interaction systems, tactile feedback devices, display screens, and virtual reality interaction interfaces. Customized training courses, case studies, and simulation exercises are delivered based on the trainees' training progress and assessment results. S6. Based on the multi-dimensional evaluation data and feedback information of students' multiple training sessions, a personalized learning path planning algorithm is used to develop an exclusive training plan for each student.

2. The method for intelligently assisting helicopter pilots in driving skills training according to claim 1, characterized in that: The sensors include a displacement sensor arranged on the operating lever, a pressure sensor arranged on the foot pedal, a posture sensor arranged on the seat, an eye tracking sensor, an electroencephalogram sensor, a heart rate sensor and a skin charge sensor.

3. The method for intelligently assisting helicopter pilots in driving skills training according to claim 2, characterized in that: The EEG sensor collects electrical activity signals from the student's brain to analyze the student's cognitive load and concentration level; The heart rate sensor monitors the student's heart rate changes in real time, reflecting the student's emotional state and physical stress level; the skin conductivity sensor detects the student's skin conductivity to assist in judging the student's emotional fluctuations.

4. The intelligent auxiliary training method for helicopter pilot driving skills according to claim 3 is characterized by: In step S2, based on the comprehensive data model, the deviation value of the operation index is calculated, and the correlation between the trainee's physiological and psychological state and the operation performance is analyzed.

5. The method for intelligently assisting helicopter pilots in driving skills training according to claim 4, characterized in that: In step S3, if the trainee performs poorly on a specific operation, hovering training scenarios under different wind speeds and terrain conditions are generated; if the trainee is under high stress, a relatively simple and relaxing flight scenario is generated to relieve stress, and prompt elements are set in the scenario to gradually guide the trainee to correct operation.

6. The intelligent auxiliary training method for helicopter pilot driving skills according to claim 5, characterized in that: In step S4, the comprehensive assessment results include evaluations of operational skills, adaptability, and psychological stability.

7. The intelligent auxiliary training method for helicopter pilot driving skills according to claim 6, characterized in that: In step S5, the voice interaction system provides real-time conversational guidance based on the evaluation results; The tactile feedback device simulates abnormal conditions during helicopter flight through vibrations of different intensities and frequencies to remind trainees; the display screen and virtual reality interactive interface display operational deviations and improvement suggestions in the form of three-dimensional animations and highlighted labels.

8. The intelligent auxiliary training method for helicopter pilot driving skills according to claim 7, characterized in that: The exclusive training plan dynamically adjusts the difficulty of training scenarios, the focus of training content, and the training schedule; regularly reviews and analyzes the training results of students, continuously optimizes the training plan, and synchronizes relevant information to the coaching terminal.

9. An intelligent auxiliary training system for helicopter pilot driving skills, used to implement the intelligent auxiliary training method for helicopter pilot driving skills according to any one of claims 1 to 8, characterized in that: include: Multi-source data acquisition module: including operational sensors, physiological sensors, and eye tracking sensors, used to collect various data of trainees during training in real time; Data fusion and processing module: Receives data from the multi-source data acquisition module, performs pre-processing and deep fusion, and performs analysis and calculation to build a comprehensive data model of the trainee's training status; Dynamic scene generation module: Based on the results of the data fusion and processing module, it uses VR and AR to dynamically generate diverse and personalized simulated flight scenes; Intelligent multimodal assessment module: uses a deep learning multimodal assessment model to comprehensively evaluate students' performance in dynamic scenarios, and outputs multi-dimensional assessment results and detailed reports; Real-time interactive feedback module: used to interact with trainees, provide guidance and push training resources; Personalized training optimization module: Develop and adjust personalized training plans based on trainee training data, continuously optimize the training process, and interact with the coach's terminal for data; Coach terminal: used for coaches to view students' training data, evaluation reports and training plans in real time, provide remote guidance, intervene in the training process, communicate with students, and provide personalized training suggestions.

Images