Flight training operation efficiency evaluation method based on flight operation data

Abstract

The invention relates to the technical field of flight training effect evaluation and data analysis, in particular to a flight training operation efficiency evaluation method based on flight operation data, and the method comprises the steps: collecting multi-source data in the flight training of a trainee, and segmenting the multi-source data into structured training segments according to subjects; evaluating the intervention effect of the instructor intervention event in each segment, and calculating the comprehensive score of the teaching interaction efficiency; calculating a comprehensive skill score of each training of the trainee in a specific subject, establishing a personal skill attenuation model, and predicting a future skill gap; analyzing the association between the teaching interaction efficiency score and the skill attenuation rate; a real-time feedback report and a personalized flight training plan are generated based on the results, and fine management of the teaching process is realized by quantifying the relationship between the intervention effect and skill attenuation, so that the pertinence of instructor guidance and the skill keeping efficiency of trainees are improved, the training resource configuration is optimized, and the scientificity and personalized level of flight training are enhanced.

Description

Technical Field

[0001] This invention relates to the field of flight training effectiveness evaluation and data analysis technology, specifically to a method for evaluating flight training operation efficiency based on flight operation data. Background Technology

[0002] Current flight training effectiveness evaluations rely heavily on instructors' subjective experience and lack systematic quantitative analysis of multi-source operational data. This makes it difficult to accurately assess the relationship between teaching interaction efficiency and student skill attenuation, resulting in insufficient targeting of training plans and low resource utilization efficiency. This invention constructs an objective evaluation system based on flight operational data to improve training efficiency and personalized teaching levels.

[0003] Existing or traditional methods for evaluating the efficiency of flight training operations based on flight operation data have at least the following technical problems: 1. Traditional flight training operation efficiency evaluation methods based on flight operation data lack a deep integration and structured processing mechanism for multi-source heterogeneous data. They often focus on the ex-post analysis of single-dimensional flight parameters (such as attitude, speed and altitude) or only conduct independent review of voice data. The lack of automated alignment and synchronization between data based on accurate timestamps makes flight control data, voice communication content and training subject events isolated on the timeline. As a result, the evaluation process cannot restore the complete closed-loop real teaching interaction scenario, and the evaluation basis remains on isolated data points.

[0004] 2. Existing flight training operation efficiency evaluation methods based on flight operation data lack a refined and quantitative evaluation framework for instructor intervention behavior. Most of them only record whether the instructor has intervened or make simple qualitative classifications of the intervention (such as correction or prompting), lacking quantitative analysis of the key attributes of the intervention, which leads to the evaluation of instructor teaching behavior being subjective and general.

[0005] 3. Traditional flight training operation efficiency evaluation methods based on flight operation data lack the ability to model and predict the dynamic decay process of trainees' individual skill status. They usually use static "pass / fail" or average scores as the main conclusions, and lack the ability to build personalized skill decay models based on trainees' historical training sequence data (a sequence of comprehensive skill scores changing over time). Since the specific rate of skill decay as training intervals increase is not quantified, it is impossible to predict the skill gap that trainees may have at a certain point in the future. As a result, training arrangements mainly rely on the fixed cycle of the syllabus or instructor experience, and cannot achieve accurate and preventive planning based on the trainees' individual forgetting curves.

[0006] 4. Traditional flight training operation efficiency evaluation methods based on flight operation data lack empirical analysis of the causal or correlation relationship between teaching process efficiency and long-term training effects. Traditional methods treat the evaluation of a single training session and the long-term progress of trainees as relatively independent parts, lacking a systematic exploration of whether there is a quantifiable influence relationship between the efficiency score of teaching interaction in each training session and the rate of skill decay of trainees after the end of this training session and before the start of the next training session. Due to the lack of such correlation analysis, it is impossible to verify from the data which teaching methods and intervention strategies are more conducive to slowing down skill forgetting and consolidating long-term learning effects, thus resulting in a lack of data-driven direction for teaching improvement. Summary of the Invention

[0007] To address the aforementioned shortcomings of existing technologies, this invention provides a flight training operation efficiency evaluation method based on flight operation data, which can effectively solve the problems of subjective experience-driven, inefficient, and unpredictable training evaluation in the background technology.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for evaluating the efficiency of flight training operations based on flight operation data, comprising: S1. Collect multi-source flight data generated by trainees during each flight training session, and divide and encapsulate it into structured training segment data units according to training subjects.

[0009] The multi-source flight data includes flight control and status data, cockpit voice data, and training event tagging data.

[0010] S2. Evaluate the intervention effect of the instructor intervention events in each training segment, and calculate the comprehensive score of teaching interaction efficiency corresponding to each training segment.

[0011] S3. Calculate the student's comprehensive skill score in each training session in a specific training subject, establish a skill decay model for the student in the specific training subject, and predict the student's skill gap at any future time point.

[0012] S4. Calculate the observed skill decay rate between adjacent training sessions in a specific training subject, and evaluate the impact between the comprehensive score of teaching interaction efficiency and the skill decay rate.

[0013] S5 generates real-time feedback reports and flight training plans for trainees.

[0014] Preferably, the process of segmenting and encapsulating training segment data units according to training subjects is as follows: Based on the set recording frequency, multi-source flight data with timestamps generated by trainees during each flight training session are collected.

[0015] Using a synchronized clock as a common time reference, multi-source flight data are aligned on the time axis.

[0016] Based on the start and end timestamps of each training subject in the training event labeling data, extract the multi-source flight data corresponding to each complete training subject cycle, and encapsulate it into the corresponding structured data unit, which is recorded as the corresponding training segment.

[0017] Preferably, the specific process for evaluating the intervention effect of the instructor intervention events in each training segment is as follows: Each intervention event initiated by the instructor in each training segment is obtained, and the corresponding start and end timestamps are extracted. The duration of each intervention event is calculated by subtracting the timestamps.

[0018] The cabin voice data text corresponding to the timestamps in each training segment is classified to obtain the type of each intervention event.

[0019] Based on the type of each intervention event, the corresponding key flight parameters are obtained. Based on the preset fixed analysis window duration and the set recording frequency, the standard deviation of the key flight parameters in the pre-intervention analysis window and post-intervention analysis window for each intervention event is calculated.

[0020] By combining the duration of each intervention event in each training segment, the instantaneous quantitative value of each intervention event in each training segment is calculated.

[0021] The duration during which all key flight parameters are simultaneously within the corresponding preset range is statistically analyzed. Combined with the preset autonomous observation duration for each training segment, the quantitative value of the autonomy maintenance index corresponding to each intervention event in each training segment is calculated. This allows for the evaluation of the intervention effect of instructor intervention events in each training segment.

[0022] Preferably, the process for calculating the comprehensive score of teaching interaction efficiency for each training segment is as follows: Obtain the intervention delay duration for each intervention event corresponding to each training segment, and calculate the average intervention delay duration for each training segment.

[0023] Based on the instantaneous indicator quantification value and the autonomy maintenance indicator quantification value corresponding to each intervention event in each training segment, combined with the pre-set intervention delay penalty coefficient, and the weight factors corresponding to the instantaneous indicator quantification value and the autonomy maintenance indicator quantification value, the comprehensive score of teaching interaction efficiency corresponding to each training segment is calculated.

[0024] Preferably, the calculation of the trainee's comprehensive skill score in each training session of a specific training subject is carried out as follows: Obtain historical training segments for each student's specific training subject, extract key performance indicators (KPIs) from them, and select the KPIs corresponding to each training session.

[0025] Based on a pre-defined standardization function, each key performance indicator is mapped to a corresponding interval. Combining the pre-defined standardization function and the weight coefficients of each key performance indicator for each training session, the comprehensive skill score of the trainee for each training session for a specific training subject is calculated.

[0026] Preferably, the specific process of establishing the skill decay model for trainees in a specific training subject is as follows: Obtain the comprehensive skill score sequence of trainees in a specific training subject and the timestamp of each training session. Based on the comprehensive skill score sequence and the corresponding timestamp, fit a skill decay model of trainees in a specific training subject.

[0027] Preferably, the process of predicting the student's skill gap at any future point in time is as follows: The system obtains the date on which the trainee plans to conduct the next training session. Based on the skill decay model, it calculates the trainee's predicted skill score for a specific training subject on that training date. Combined with a pre-set skill attainment threshold, it calculates the quantitative value of the trainee's predicted skill gap for that specific training subject on that training date.

[0028] Based on the trainee's predicted skill gap quantification value for a specific training subject on the training date and the preset skill attainment threshold, assess whether the trainee has a skill gap for the specific training subject on the training date.

[0029] If it exists, obtain the student's average skill score improvement value for a single training session for a specific training subject, and combine it with the predicted skill gap quantification value to calculate the student's additional training sessions for that specific training subject on that training date.

[0030] Preferably, the process of calculating the rate of decline of observed skills between adjacent training sessions in a specific training subject is as follows: Obtain the trainee's overall skill score for adjacent training segments for a specific training subject.

[0031] Obtain the actual interval duration between adjacent training segments for a specific training subject for that student.

[0032] The rate of decline of observational skills between adjacent training segments in a specific training subject was calculated.

[0033] Preferably, the specific process for assessing the impact between the comprehensive score of teaching interaction efficiency and the skill decay rate is as follows: A related dataset is constructed based on the comprehensive score of teaching interaction efficiency of each training segment and the observed skill decay rate between adjacent training segments in a specific training subject.

[0034] Statistical analysis was performed on the associated datasets to obtain the parameters of the correlation model that quantifies the influence of the comprehensive score of teaching interaction efficiency on the rate of decline of observed skills.

[0035] Preferably, the process of generating the real-time feedback report and flight training plan corresponding to the trainee is as follows: Based on the comprehensive score of teaching interaction efficiency, the individual skill decay model, and the correlation model, optimization strategies are generated, including real-time teaching feedback reports and flight training plans.

[0036] Based on the comprehensive score of teaching interaction efficiency and its sub-indicators for the current training segment of the student, and by comparing it with historical data, a real-time teaching feedback report is generated for the student.

[0037] Based on the prediction results of the skill decay model, the parameters of the correlation model, and the preset skill attainment requirements, the recommended training intervals are calculated and the training subjects that need attention are identified, thereby generating the corresponding flight training plan for the trainees.

[0038] The technical solution provided by this invention has the following advantages compared with the known prior art: 1. In the process of integrating and structuring multi-source flight data, the embodiments of the present invention utilize timeline alignment technology based on a synchronous clock and an automatic segmentation and encapsulation method based on event tagging. This facilitates the construction of complete, coherent, and machine-readable training context units. It employs industrial-grade synchronous clocks and event tagging devices to integrate flight parameters, voice streams, and teaching events into a unified training segment with millisecond-level precision. This ensures that all subsequent analyses are based on a real scenario with a complete context, rather than discrete data points, thus laying a reliable data foundation for refined evaluation.

[0039] 2. In the quantitative evaluation of the instructor intervention effect, this invention combines natural language processing text classification with statistical analysis of key flight parameters over a time window. This facilitates an objective and multi-dimensional measurement of the quality of teaching interaction. The invention uses a pre-trained language model for real-time intent classification of cabin speech and immediately associates the classification results with preset rules to dynamically select corresponding key flight parameters. Subsequently, standard statistical analysis is applied to fixed time windows before and after the intervention to quantify the immediate correction effect and the autonomy retention effect, thus realizing the concretization and data-driven evaluation of abstract teaching skills.

[0040] 3. In the process of modeling and predicting individual student skills, this invention employs a method based on standardized function mapping for calculating comprehensive skill scores and a nonlinear least squares fitting exponential decay model. This facilitates highly personalized skill status diagnosis and proactive early warning. The invention utilizes the membership function concept from fuzzy set theory as a standardized function, uniformly mapping flight performance indicators of different dimensions and meanings into dimensionless skill score components. A weighted comprehensive score is then synthesized, employing the general form of an exponential decay function describing the natural forgetting pattern. Furthermore, the existing optimization technique of nonlinear least squares is applied to fit the individual student's historical score sequence, solving for the individual's unique initial level and decay rate parameters. This transforms skill assessment from a static rating of past performance into a dynamic and continuous prediction model of an individual's future skill status.

[0041] 4. In the process of analyzing the correlation between teaching efficiency and long-term effects, this invention establishes a linear regression statistical model between the comprehensive score of teaching interaction efficiency and the observed skill decay rate. This model helps to reveal the quantitative relationship between teaching behavior and skill retention, scientifically linking the teaching process with long-term effects. It also applies the classic statistical tool of linear regression analysis. The innovation lies in the selected variables: the comprehensive score of teaching interaction efficiency, which quantifies the quality of a single teaching session, is used as the independent variable, and the observed skill decay rate, which reflects the speed of long-term forgetting, is used as the dependent variable. The slope and intercept are fitted using the least squares method. This model helps to directly and quantitatively answer the core question: "How much daily skill forgetting is expected to be slowed down by increasing the teaching efficiency score by one unit?" This application elevates statistical methods from general correlation analysis to a medium for causal reasoning that connects micro-level training operations with macro-level effectiveness.

[0042] 5. In the personalized training decision support generation process, the embodiments of the present invention adopt a strategy generation logic that integrates teaching efficiency scores, personal skill decay models and related model parameters, which is conducive to outputting accurate decision suggestions that have both real-time guidance and long-term planning. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0044] Figure 1 This is a schematic diagram showing the connection of the method steps of the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0046] The present invention will be further described below with reference to embodiments.

[0047] Please see Figure 1 As shown, the flight training operation efficiency evaluation method based on flight operation data includes at least the following: S1. Collect multi-source flight data generated by trainees during each flight training session, and divide and encapsulate it into structured training segment data units according to training subjects.

[0048] In a specific embodiment, the process of segmenting and encapsulating training segment data units according to training subjects is as follows: the multi-source flight data includes flight control and status data, cockpit voice data, and training event tagging data.

[0049] Flight control and status data includes flight control input data and aircraft response status. Based on a set recording frequency, the flight control input data and aircraft response status at each time point corresponding to each flight training session are collected and recorded by airborne data recording equipment. Flight control input data includes, but is not limited to, stick displacement and rudder pedal displacement. Aircraft response status includes, but is not limited to, attitude angle, airspeed, altitude, and heading. This generates timestamped flight control and status data.

[0050] The cockpit audio management system collects and distinguishes the audio channels of instructors and trainees. Through an automatic speech recognition engine, the collected audio of instructors and trainees is converted into time-stamped interactive text, thereby generating time-stamped cockpit voice data.

[0051] Using specialized recording equipment, instructors record the start and end events of each flight training session and mark them with corresponding timestamps, thereby generating training event marker data with timestamps.

[0052] Using a synchronized clock as a common time reference, the flight control and status data, cockpit voice data, and training event marker data are aligned on the timeline. Based on the start and end timestamps of each training subject in the training event marker data, the flight control and status data, cockpit voice data, and training event marker data corresponding to each complete training subject cycle are extracted from the aligned flight control and status data, cockpit voice data, and training event marker data, and encapsulated into corresponding structured data units, which are recorded as the corresponding training segments.

[0053] It should be noted that the recording frequency is determined based on the physical performance of the airborne data recording equipment and the typical time scale of the flight control maneuvers to be analyzed, and is usually set to no less than 1Hz.

[0054] It should be noted that the cockpit audio management system is a standard component of the existing aircraft communication system, referring to the hardware and software integration of the cockpit voice recorder (CVR) or dedicated training audio acquisition equipment. In this solution, it is used to automatically separate and record the independent audio channels of instructors and trainees according to preset rules (such as based on physical microphone position or voiceprint characteristics), providing a structured raw data source for subsequent voice analysis.

[0055] It should be noted that the automatic speech recognition engine is based on existing automatic speech recognition technology and can adopt mature cloud or local ASR services / engines. In this solution, its core function is as a data processing tool to convert the continuous audio stream collected in the previous step into a structured dialogue text sequence with millisecond-level timestamps in real time or offline. This transforms unstructured speech information into machine-readable text that can be aligned and correlated with the flight parameter timeline.

[0056] It should be noted that dedicated recording devices refer to specific hardware terminals or software applications integrated into the training management system, used by instructors to record boundary events of training subjects and automatically generate precise timestamps synchronized with the system's master clock.

[0057] It should be noted that a synchronization clock provides a unified, stable, and less-than-millisecond-level common time source for all multi-source flight data acquisition devices, such as a GPS clock or a network time protocol server, to ensure that multi-source flight data can be accurately aligned based on the same timeline.

[0058] In the process of integrating and structuring multi-source flight data, this invention employs timeline alignment technology based on a synchronous clock and an automatic segmentation and encapsulation method based on event tagging. This facilitates the construction of complete, coherent, and machine-readable training context units. It utilizes industrial-grade synchronous clocks and event tagging equipment to fuse flight parameters, voice streams, and teaching events into a unified training segment with millisecond-level precision. This ensures that all subsequent analyses are based on a contextually complete real-world scenario, rather than discrete data points, thus laying a reliable data foundation for refined evaluation.

[0059] S2. Based on the generated training segments, evaluate the intervention effect of the teacher intervention events in each training segment, and calculate the comprehensive score of teaching interaction efficiency corresponding to each training segment.

[0060] In one specific embodiment, the evaluation of the intervention effect of instructor intervention events in each training segment is carried out as follows: cabin voice data with corresponding timestamps are obtained from each training segment, each intervention event initiated by the instructor in each training segment is identified, the corresponding start timestamp and end timestamp are extracted from each intervention event in each training segment, and the duration of each intervention event is obtained by subtraction.

[0061] By using a natural language processing model, the cabin voice data text with corresponding timestamps in each training segment is classified to obtain the type of each intervention event.

[0062] Based on the type of each intervention event, corresponding key flight parameters are obtained from flight control and status data. Using a pre-set fixed analysis window duration and recording frequency, the standard deviations of the key flight parameters within the pre-intervention and post-intervention analysis windows for each intervention event are calculated and denoted as follows: and .

[0063] Calculation formula: Calculate the first In the training segment, the first The immediate quantitative value of the indicator corresponding to the intervention event ,in The numbers corresponding to each training segment, The value of is a positive integer. Each intervention event is assigned a number. The value of is a positive integer. Represented as the first In the training segment, the first Duration of each intervention event.

[0064] Based on the key flight parameters matched to each intervention event for each training segment, and using pre-defined acceptable ranges for these key flight parameters, logical AND rules are employed to calculate the total acceptable time for each key flight parameter to be simultaneously within its corresponding pre-defined acceptable range. Combined with the pre-set autonomous observation time corresponding to each training segment Through the calculation formula: , obtained the In the training segment, the first Quantitative value of autonomy maintenance index corresponding to the intervention event .

[0065] It should be noted that the intervention event refers to a single complete instruction statement issued by the instructor to the trainee via voice during flight training, which is intended to correct operation or guide attention. It is the basic unit for subsequent quantitative analysis. For example, in a "non-precision approach" training segment, the system recognizes from the cockpit voice data that the instructor says "Pay attention to heading, correct to the right by 5 degrees". This sentence from beginning to end is a specific intervention event.

[0066] It should be noted that the natural language processing model refers to a text classifier built based on existing pre-trained language models (such as BERT and RoBERTa). In this scheme, the model is used to classify the intent of the text converted from speech recognition and automatically map it to a preset set of intervention types (such as "guiding questions", "corrective instructions" and "encouraging feedback"). This classification result is the core basis for the subsequent quantitative analysis to associate key flight parameters and set evaluation rules (such as autonomous observation duration).

[0067] It should be noted that the fixed analysis window duration is set in advance based on the typical operational cycle and assessment requirements of flight training subjects, usually between 30 and 60 seconds, to ensure that the complete operational evolution process before and after the intervention is covered.

[0068] It should be noted that the pre-intervention analysis window refers to the time interval formed by tracing back a preset fixed analysis window duration (e.g., 30 seconds) from the start of an intervention event, used to analyze the trainee's operational status before the intervention; the post-intervention analysis window refers to the time interval formed by extending forward a preset fixed analysis window duration from the end of the intervention event, used to analyze the trainee's operational response and stability after the intervention.

[0069] It should be noted that key flight parameters refer to flight control or state variables that are directly related to the semantic content of a specific intervention event and are used to quantitatively evaluate the effectiveness of the intervention. For example, for interventions involving attention airspeed, the key flight parameter is indicated airspeed; for interventions involving maintaining heading, the key flight parameter is heading deviation. The process for setting the acceptable range of key flight parameters is as follows: the range is a pre-set quantitative threshold based on the operational specifications in the flight manual, the scoring standards in the training syllabus, and the statistical distribution of historical safe operation data for the aircraft type. For example, for the "landing slope" parameter, its acceptable range can be set to [-2 degrees, +2 degrees] based on the landing performance standards in the flight manual.

[0070] In one specific embodiment, the process of calculating the comprehensive score of teaching interaction efficiency corresponding to each training segment is as follows: The intervention delay duration for each intervention event corresponding to each training segment is obtained, and the average intervention delay duration corresponding to each training segment is calculated by averaging. Based on the real-time quantitative values ​​of the indicators and the quantitative values ​​of the autonomy maintenance indicators corresponding to each intervention event in each training segment, combined with the pre-set intervention delay penalty coefficient, Through the calculation formula: , obtained the The overall score of teaching interaction efficiency corresponding to each training segment and instructor intervention event. ,in , This represents the total number of corresponding intervention events in the training segment. The value of is a positive integer. and These represent the weight factors corresponding to the quantitative values ​​of the real-time indicators and the weight factors corresponding to the quantitative values ​​of the autonomy maintenance indicators, respectively.

[0071] It should be noted that the specific process for obtaining the intervention delay duration for each training segment corresponding to each intervention event is as follows: based on the preset qualified range of key flight parameters and the set recording frequency, the moment when the key flight parameters exceed the qualified range of key flight parameters during the trainee's operation is detected and recorded as the deviation occurrence timestamp. The difference between the deviation occurrence timestamp and the start timestamp of the corresponding intervention event is used to obtain the single intervention delay. All timestamps are based on the synchronization clock reference of step S1.

[0072] It should be noted that the process of setting the intervention delay penalty coefficient is as follows: by analyzing historical training data, the quantitative relationship between intervention delay and skill recovery cost is determined (for example, statistics show that for every 1 second increase in average delay, the subsequent skill consolidation cost increases by about 2%), thereby transforming the impact of such business into a specific mathematical correction coefficient (e.g., 0.02), so that the comprehensive score of teaching interaction efficiency can intuitively reflect the efficiency loss caused by the delay.

[0073] It should be noted that, and The values ​​are greater than 0 and less than 1. and The sum of all is always 1. and The setting is based on the analysis of historical training data in this scheme. By collecting intervention events under different training subjects in historical data, the immediate quantitative values ​​of the indicators and the quantitative values ​​of the autonomy maintenance indicators are calculated. Statistical regression analysis is used to analyze the correlation between these indicators and the subsequent skill decay rate of the corresponding trainees to determine the relative contribution of the two to long-term skill retention. Finally, the contribution ratio is normalized and set as follows: and (For example, if the analysis shows that immediate correction contributes approximately 60% and autonomous cultivation contributes approximately 40%, then set...) It is 0.6. (0.4).

[0074] In the quantitative evaluation of instructor intervention effects, this invention combines natural language processing text classification with statistical analysis of key flight parameters over time windows. This approach facilitates an objective and multi-dimensional measurement of the quality of teaching interactions. The invention uses a pre-trained language model for real-time intent classification of cabin speech and immediately associates the classification results with preset rules to dynamically select corresponding key flight parameters. Subsequently, standard statistical analysis is applied over fixed time windows before and after intervention to quantify the effects of immediate correction and maintenance of autonomy. This achieves a concrete and data-driven evaluation of abstract teaching techniques.

[0075] S3. Based on the trainee's historical training segment sequence in a specific training subject, calculate the trainee's comprehensive skill score for each training session in the specific training subject, and establish a skill decay model for the trainee in the specific training subject to predict the trainee's skill gap at any future time.

[0076] In one specific embodiment, the calculation of a student's comprehensive skill score for each training session in a specific training subject involves the following steps: Based on the specific training subject corresponding to the student, historical training segments for that subject are retrieved from the database, and key performance indicators (KPIs) are extracted from them. For each training session in the specific training subject, the corresponding KPIs are selected and denoted as follows: ,in These are the numbers for each key performance indicator. , This represents the total number of key performance indicators (KPIs) during each training session. The value of is a positive integer. The corresponding number for each training session. The value can be a positive integer.

[0077] Through a pre-defined standardized function Each key performance indicator is mapped to the range [0, 100]. Represented as the first In the training session, the first The specific values ​​of each key performance indicator are entered up to the [number]th [key performance indicator]. The standardized function preset for each key performance indicator is used to calculate the first key performance indicator. The standardized scores of the key performance indicators measured in this study.

[0078] Calculation formula: The student receives the first result for the specific training subject. The overall skill score corresponding to this training session , For the set number The training corresponds to the first The weighting coefficients of each key performance indicator.

[0079] It should be noted that specific training subjects, such as crosswind landing and non-precision approach, have clearly defined operating procedures and evaluation standards for flight training units; key performance indicators, such as longitudinal deviation of landing touch point, peak vertical overload, and glide path deviation, are directly extracted from multi-source flight data.

[0080] It should be noted that the standardization function is based on existing data normalization methods and the concept of membership function in fuzzy set theory. In this scheme, it is used to transform key performance indicators with different physical dimensions and numerical ranges into unified, weighted, and summable dimensionless skill score components. Specifically, it takes the form of a parameterized function selected based on the characteristics of the indicator. For example, for indicators with well-defined ideal values ​​and symmetrical deviations, a Gaussian function is used. , Represented as the first The preset ideal values ​​for each key performance indicator Indicated for the first The width parameter set by the key performance indicators to control the decay rate of the normalized function output. Represented as standardized coefficients, due to the use of a pre-defined standardized function. Mapping each key performance indicator to the range [0, 100], then here... The value is 100.

[0081] It should also be noted that, The value is determined directly based on the standard operating procedures or optimal theoretical values ​​in existing technical specifications. For example, for the indicator "longitudinal deviation of the landing point," its ideal value... Setting it to 0 meters represents a perfect landing point; for "glide path tracking deviation during non-precision approaches," its Similarly, setting it to 0 points indicates that it completely follows the standard glide path.

[0082] Calibration is achieved by comprehensively analyzing historical data on "qualified" operational performance and expert experience thresholds. For example, for "longitudinal deviation of the landing touchpoint," the standard deviation of this indicator is first statistically analyzed in all historically evaluated "qualified" landing records (e.g., calculated to be 15 meters). This is then combined with the available runway length in the flight manual and the generally accepted tolerance range for instructors (e.g., ±20 meters) to ultimately determine the deviation. Set a composite value (e.g., 18 meters) to quantitatively define the deviation boundary where the score drops significantly from the maximum score. It should also be noted that, The values ​​are all greater than 0 and less than 1. The setting process is as follows: Based on the historical training data analysis process of this scheme, the correlation between the historical data of each key performance indicator and the subsequent long-term skill maintenance level of trainees is analyzed through statistical regression method. In this way, the contribution of each indicator to the prediction of the overall skill status is quantified, and the contribution is normalized and set as the weight of each indicator (for example, if the contribution of the three indicators of ground position deviation, vertical overload and heading maintenance are 50%, 30% and 20% respectively, then their weights are set to 0.5, 0.3 and 0.2 respectively).

[0083] In one specific embodiment, the process of establishing a skill decay model for trainees in a specific training subject is as follows: Obtain the trainee's comprehensive skill score in the specific training subject, sorted by training time, and the corresponding time points for each training session. Then, use the nonlinear least squares method to fit the following skill decay model, which takes the form: ,in This represents the trainees at the predicted time point. Skill scores for specific training subjects This indicates the student's initial skill level score. It is a natural constant, approximately equal to 2.71828. It is expressed as the skill decay rate constant.

[0084] It should be noted that the process of fitting the model using the nonlinear least squares method is based on the parameter estimation technique in the existing mathematical optimization theory. The model is in the form of the exponential function that describes the natural decay or forgetting phenomenon. In this scheme, it is used to fit the discrete historical skill score data points of the trainees into a continuous personalized skill decay curve, thereby quantifying their skill decay rate.

[0085] It should be noted that, and The acquisition process is as follows: Based on the trainee's historical training data in a specific training subject, including the timestamps of each historical training session and the historical comprehensive skill score, the results are obtained through reverse engineering using a model. and The value, for example, during the fitting process, the computer will use optimization algorithms such as Levenberg-Marquardt to automatically iterate and solve the problem, first guessing a set of initial parameters (e.g. It is 100. The algorithm first sets the value to 0.01, then calculates the predicted values ​​for all historical data points (e.g., 5 points) under this parameter, and summarizes the total error (sum of squared residuals) between the predicted and actual values. Next, the algorithm intelligently adjusts its parameters in directions that rapidly reduce the total error. and The value (e.g., adjusted to) It is 98.2. The error is calculated repeatedly (to a value of 0.0152). After multiple iterations, the algorithm stops when the total error can no longer be significantly reduced. At this point, the result is... and This is the optimal solution.

[0086] In one specific embodiment, the process of predicting the trainee's skill gap at any future point in time is as follows: Obtain the date the trainee plans to attend their next training session. The date is input into the skill decay model to calculate the trainee's predicted skill score for a specific training subject on that training date. Combined with pre-set skill attainment thresholds Through the calculation formula: The trainees were informed on the training date. Quantitative values ​​of predicted skill gaps for specific training subjects .

[0087] when ≥ hour, A value of 0 indicates that the trainee was on that training date. For specific training subjects, the predicted skill achievement level, when < hour, >0 indicates that the trainee was on that training date. There are skill gaps in specific training subjects, requiring additional training.

[0088] The student's historical training data for a specific subject was retrieved from the database, and the average skill score improvement per training session was calculated. , combined Through the calculation formula: The trainee was obtained on the training date. Additional training sessions for specific training subjects .

[0089] It should be noted that the system retrieves all historical training segments of the student in the specific training subject from the database, reads the comprehensive skill score of each segment calculated by step S3, then calculates the change in skill score between two adjacent training sessions, and filters out data that are greater than 0 (i.e., the performance in the later training session is better than the previous one). Finally, the system calculates the arithmetic mean of all positive changes to obtain the student's average skill improvement value in a single training session in this subject.

[0090] It should be noted that the process of setting the pre-set skill attainment threshold integrates objective standards and historical data: First, the basic passing line is determined based on the flight manual specifications, standard operating procedures, and clear scoring criteria in the training syllabus for this training subject. Second, all training segments in the history of this aircraft type or training institution in which trainees have been rated as qualified or passed in this subject are analyzed, and the median or specific percentile of their comprehensive skill scores is calculated. Finally, the above two are calibrated in combination with expert experience to obtain the set skill attainment threshold.

[0091] In the process of modeling and predicting individual student skills, this invention employs a method based on standardized function mapping for calculating comprehensive skill scores and using a nonlinear least squares fitting exponential decay model. This facilitates highly personalized skill status diagnosis and proactive early warning. The invention utilizes the membership function concept from fuzzy set theory as a standardized function, mapping flight performance indicators of different dimensions and meanings into dimensionless skill score components. A weighted composite score is then generated, employing the general form of an exponential decay function describing the natural forgetting process. Furthermore, the invention applies the existing optimization technique of nonlinear least squares to fit the individual student's historical score sequence, solving for the individual's unique initial level and decay rate parameters. This transforms skill assessment from a static rating of past performance into a dynamic and continuous prediction model for an individual's future skill status.

[0092] S4. Based on the comprehensive score of teaching interaction efficiency corresponding to each training segment and the comprehensive skill score of the trainee in each training session in a specific training subject, calculate the observed skill decay rate between adjacent training sessions in a specific training subject, and evaluate the influence between the comprehensive score of teaching interaction efficiency and the skill decay rate by establishing a statistical model.

[0093] In one specific embodiment, the calculation of the observed skill decay rate between adjacent training sessions in a specific training subject is carried out as follows: Based on the student's comprehensive skill score for each training session in the specific training subject, the student's first skill score for the specific training subject is obtained. The training segment and the first training segment The comprehensive skill score corresponding to each training segment is denoted as follows: and Get the corresponding number for the student's specific training subject. The training segment and the first training segment The actual interval between training segments is denoted as . Through the calculation formula: Calculate the student's adjacent first position in a specific training subject. The training segment and the first training segment The observation skill decay rate between training segments .

[0094] It should be noted that a specific training subject is the target unit of analysis, and each training session is an independent instance of that subject being performed by the trainee at different times; while each training segment is a structured data package for subsequent quantitative analysis, which is formed by encapsulating all the multimodal data (flight data, voice, event markers) generated in each training instance through the data fusion processing in step S1. That is, a subject contains multiple training sessions, and each training session generates a unique training segment.

[0095] In one specific embodiment, the process of evaluating the impact between the comprehensive teaching interaction efficiency score and the skill decay rate by establishing a statistical model is as follows: Retrieve consecutive historical training segments of the student for a specific training subject from the database, and then obtain the comprehensive teaching interaction efficiency score of the student for each training segment corresponding to the specific training subject. ,Will and Pair them up to form related data points Iterate through all consecutive pairs of historical training segments to construct an associated dataset.

[0096] Based on the comprehensive score of teaching interaction efficiency corresponding to each training segment and the observed skill decay rate between adjacent training segments in a specific training subject, a linear regression model is used to represent the relationship between teaching interaction efficiency and skill decay rate. The linear regression model (i.e., the correlation model) is in the following form: ,in and These are respectively represented by the rate of decline in observation skills and the comprehensive score of teaching interaction efficiency. and These represent the intercept and slope of the linear regression model, respectively.

[0097] Based on the associated dataset of observational skill decay rate and teaching interaction efficiency combined score, the least squares method is used to fit the associated dataset, and the parameter estimates that minimize the prediction error of the linear regression model are obtained. and This leads to the obtained fitted linear regression model. This is used to quantify the impact of the comprehensive score of teaching interaction efficiency on the rate of decline of observation skills.

[0098] It should be noted that, The calculation formula is based on the discrete difference form of the differential equation of the existing exponential decay mathematical model, which is a well-known mathematical calculation method. The innovation of this scheme lies in the fact that it is the first time that this formula is systematically applied to the comprehensive skill score sequence in the S3 step to calculate the rate of change of micro skills between any two adjacent training sessions (i.e. training segments) in a specific flight training subject, thereby generating basic data that can be used for fine correlation analysis with the teaching interaction efficiency score of the S2 step.

[0099] It should be noted that the linear regression model is based on the classic parametric correlation analysis method in statistics. In this scheme, it is creatively applied to establish a quantitative mathematical model between the comprehensive score of teaching interaction efficiency with S2 steps and the observed skill decay rate calculated by the preceding process with S4 steps.

[0100] It should be noted that the intercept of the linear regression model means the natural daily rate of skill decay during subsequent training intervals when the overall score of teaching interaction efficiency is zero. It quantifies the baseline forgetting rate when effective teaching is lacking. The slope of the linear regression model means the change in the daily rate of skill decay caused by each unit increase in the teaching interaction efficiency score (expected to be negative). It directly and quantitatively answers the extent to which improving teaching efficiency can slow down skill forgetting and is the core quantitative indicator linking process efficiency and long-term effects in this scheme.

[0101] It should be noted that the process of fitting the associated dataset using the least squares method is based on the classical parameter estimation method in mathematical optimization theory, used to solve for the parameters in the linear regression model. and The aim is to determine a mathematical expression of the quantitative relationship between the comprehensive score of teaching interaction efficiency and the rate of decline of observed skills from historical data.

[0102] In the process of analyzing the correlation between teaching efficiency and long-term effects, this invention establishes a linear regression statistical model between the comprehensive score of teaching interaction efficiency and the observed skill decay rate. This model helps to reveal the quantitative relationship between teaching behavior and skill retention, scientifically linking the teaching process with long-term effects. It also applies the classic statistical tool of linear regression analysis. The innovation lies in the selected variables: the comprehensive score of teaching interaction efficiency, which quantifies the quality of a single teaching session, is used as the independent variable, and the observed skill decay rate, which reflects the speed of long-term forgetting, is used as the dependent variable. The slope and intercept are fitted using the least squares method. This model helps to directly and quantitatively answer the core question: "How much daily skill forgetting is expected to be slowed down by increasing the teaching efficiency score by one unit?" This application elevates statistical methods from general correlation analysis to a medium for causal reasoning that connects micro-level training operations with macro-level effects.

[0103] S5. Generate real-time feedback reports for improving instruction and flight training plans for optimizing resource planning.

[0104] In one specific embodiment, the process of generating real-time feedback reports for improving teaching and flight training plans for optimizing resource planning is as follows: Based on the comprehensive score of teaching interaction efficiency, the personal skill decay model, and the correlation model, optimization strategies are generated, including real-time teaching feedback reports and flight training plans.

[0105] Based on the comprehensive score of teaching interaction efficiency of the current training segment and its sub-indicators (instantaneous indicator quantification value and autonomy maintenance indicator quantification value), a real-time teaching feedback report is generated for the corresponding student by comparing with historical data.

[0106] Based on the prediction results of the skill decay model, the parameters of the correlation model, and the preset skill attainment requirements, the recommended training intervals are calculated and the training subjects that need attention are identified, thereby generating the corresponding flight training plan for the trainees.

[0107] It should be noted that comparing historical data refers to comparing the various quantitative indicators generated in this training (such as the comprehensive score of teaching interaction efficiency and the comprehensive skill score) with two sets of benchmark data: one is the individual performance record and trend formed by the trainee in all past training in the same training subject (individual baseline); the other is the general performance level of all trainees or groups of trainees of the same type in the same subject (group baseline).

[0108] It should be noted that the parameters of the joint model refer to the model coefficients obtained through linear regression analysis in step S4, including the intercept and slope; the preset skill attainment requirements refer to the passing score that trainees must achieve for each specific training subject. The setting process is to comprehensively consider the official training syllabus standards of the subject, the statistical distribution of skill scores of qualified trainees in historical training data (such as the median), and determine a fixed value after review and calibration by domain experts, which serves as an objective and unified quantitative benchmark for measuring whether skills have been met.

[0109] It should be noted that the specific process of calculating the recommended interval and identifying subjects is as follows: The aforementioned model is used for decision-making calculations. First, based on the correlation model, the teaching efficiency target value required to ensure that the skill level is not lower than the target threshold at the end of the interval is deduced. Then, this skill decay rate constant and the student's current initial skill level score are substituted into their personal decay model to solve for the maximum allowable interval (through mathematical derivation (i.e., calculating the time required to decay from the starting point to the end point at a given rate) can directly calculate the longest allowable time interval between two training sessions while ensuring the target is met). This is the recommended training interval. Simultaneously, the predicted skill gap quantification value obtained from step S3 is checked. If the predicted skill gap quantification value of a certain training subject is greater than 0, it indicates that the current predicted skill level is not met, and thus the training subject is identified as a training subject requiring attention.

[0110] It should be noted that the real-time feedback report includes a quantitative evaluation summary of the teaching process (such as a comprehensive score of teaching interaction efficiency and average intervention delay) and improvement suggestions (such as recommending alternative strategies for a certain type of inefficient intervention method); the flight training plan includes recommended training intervals for different training subjects (such as recommending that a certain training subject be retrained within 7 days) and a list of subjects that are identified as needing priority in the allocation of training resources.

[0111] In the personalized training decision support generation process, this invention adopts a strategy generation logic that integrates teaching efficiency scores, personal skill decay models, and correlation model parameters, which is conducive to outputting accurate decision suggestions that combine real-time guidance and long-term planning.

[0112] The present invention also provides an evaluation model for flight operation efficiency, as detailed below: P1. Obtain the flight training operation data of a student within the set flight training cycle from the flight operation record table. The flight training operation data includes at least the planned flight duration, actual flight duration, number of flight missions, number of flight sorties, number of training resources, actual resource occupation time, and total duration of the statistical period.

[0113] P2. Divide the student's actual flight time within the set flight training period by the planned flight time to obtain the student's flight time utilization rate within the set flight training period. .

[0114] P3. Divide the number of flights taken by the trainee within the set flight training period by the duration of the set flight training period to obtain the trainee's flight activity density within the set flight training period. .

[0115] P4. Divide the number of flight training tasks completed by the student within the set flight training cycle by the average number of training tasks completed per flight to obtain the student's task completion efficiency within the set flight training cycle. .

[0116] P5. Based on the amount of training resources available to the student within the designated flight training cycle. The cumulative time spent using various resources within a set flight training cycle. And set the duration of the corresponding flight training cycle. Through the calculation formula: The resource utilization rate of the trainee within the set flight training cycle was calculated. ,in The corresponding IDs for each type of training resource. , This refers to the total number of different types of training resources available to the trainee within a given flight training cycle. The value of is a positive integer. P6. Retrieve the student's flight activity density data for each set flight training cycle during historical flight training from the database. The flight activity density data includes the historical flight activity density corresponding to each flight training session. Select the minimum and maximum flight activity densities, respectively. , By using the normalization formula: The normalized value of flight activity density was obtained. .

[0117] P7. Using time utilization rate, normalized values ​​of flight activity density, mission completion efficiency, and resource utilization rate as input parameters, a flight training operation efficiency model is developed: The flight training efficiency of the trainee within the set flight training cycle is obtained. , , , and These represent the weighting factors corresponding to time utilization rate, normalized value of flight activity density, mission completion efficiency, and resource utilization rate, respectively. , , and The values ​​of are all greater than 0 and less than 1, and satisfy . .

[0118] P8. The flight training efficiency of this student within the set flight training cycle. Compared with the set flight training operational efficiency threshold, if If the student's flight training efficiency is greater than or equal to the set threshold, it indicates that the student's flight training efficiency meets the requirements; otherwise, it does not meet the requirements.

[0119] It should be noted that the set flight training cycle is based on the stages (such as the basic flight stage and the aerobatic flight stage) or fixed duration (such as monthly or quarterly) of the flight training outline, and must ensure that the cycle contains a sufficient number of flight training tasks to support the effective calculation of various efficiency indicators.

[0120] The values ​​of the weighting factors are determined by a combination of factors, including the core objectives of flight training, statistical analysis of historical flight training data, and expert judgment.

[0121] Various training resources, including training aircraft, instructors, airborne training equipment, ground support facilities, and dedicated training airspace, directly or indirectly support flight training in various hardware, software, human resources, and airspace resources.

[0122] The established flight training operational efficiency thresholds are first determined by statistical analysis of historical operational efficiency data of similar flight trainees within the same training cycle to establish a baseline distribution range; then, in conjunction with the phased objectives of the flight training syllabus, safety operation regulations, and training resource allocation standards, specific thresholds are determined after evaluation by industry experts.

[0123] By modeling and processing the planned flight duration, actual flight duration, number of flight missions, number of flight sorties, number of training resources, actual resource occupancy time, and total duration of the statistical period directly obtained from the flight operation log, an operational efficiency model is constructed that covers time utilization, flight activity density, mission completion efficiency, and resource occupancy rate. The multi-dimensional operational parameters are transformed into a unified comprehensive efficiency index, thereby enabling objective quantitative evaluation and horizontal comparative analysis of flight training operational efficiency. This comprehensively reflects the organizational efficiency and resource utilization level of flight training operations, providing a scientific, intuitive, and quantifiable technical basis for training unit operational status assessment, operational plan optimization, and rational allocation of training resources.

[0124] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the efficiency of flight training operations based on flight operation data, characterized in that, include: S1. Collect multi-source flight data generated by trainees during each flight training session, and divide and encapsulate it into structured training segment data units according to training subjects; The multi-source flight data includes flight control and status data, cockpit voice data, and training event tagging data; S2. Evaluate the intervention effect of the instructor intervention events in each training segment, and calculate the comprehensive score of teaching interaction efficiency corresponding to each training segment; S3. Calculate the student's comprehensive skill score in each training session in a specific training subject, establish a skill decay model for the student in the specific training subject, and predict the student's skill gap at any future time point. S4. Calculate the observed skill decay rate between adjacent training sessions in a specific training subject, and evaluate the impact between the comprehensive score of teaching interaction efficiency and the skill decay rate. S5 generates real-time feedback reports and flight training plans for trainees.

2. The method for evaluating flight training operation efficiency based on flight operation data according to claim 1, characterized in that, The specific process of segmenting and encapsulating training segment data units according to training subjects is as follows: Based on the set recording frequency, multi-source flight data with timestamps generated by trainees during each flight training session are collected. Using a synchronized clock as a common time reference, multi-source flight data are aligned on the timeline; Based on the start and end timestamps of each training subject in the training event labeling data, extract the multi-source flight data corresponding to each complete training subject cycle, and encapsulate it into the corresponding structured data unit, which is recorded as the corresponding training segment.

3. The method for evaluating flight training operation efficiency based on flight operation data according to claim 2, characterized in that, The specific process for evaluating the intervention effectiveness of instructor intervention events in each training segment is as follows: Obtain each intervention event initiated by the instructor in each training segment, extract the corresponding start timestamp and end timestamp, and calculate the duration of each intervention event by subtraction; The cabin voice data text with corresponding timestamps in each training segment is classified to obtain the type of each intervention event; Based on the type of each intervention event, the corresponding key flight parameters are obtained. Based on the preset fixed analysis window duration and the set recording frequency, the standard deviation of the key flight parameters in the pre-intervention analysis window and post-intervention analysis window for each intervention event is calculated. By combining the duration of each intervention event in each training segment, the instantaneous quantitative value of each intervention event in each training segment is calculated. The duration during which all key flight parameters are simultaneously within the corresponding preset range is statistically analyzed. Combined with the preset autonomous observation duration for each training segment, the quantitative value of the autonomy maintenance index corresponding to each intervention event in each training segment is calculated. This allows for the evaluation of the intervention effect of instructor intervention events in each training segment.

4. The method for evaluating flight training operation efficiency based on flight operation data according to claim 3, characterized in that, The specific process for calculating the comprehensive score of teaching interaction efficiency for each training segment is as follows: Obtain the intervention delay duration for each intervention event corresponding to each training segment, and calculate the average intervention delay duration for each training segment; Based on the instantaneous indicator quantification value and the autonomy maintenance indicator quantification value corresponding to each intervention event in each training segment, combined with the pre-set intervention delay penalty coefficient, and the weight factors corresponding to the instantaneous indicator quantification value and the autonomy maintenance indicator quantification value, the comprehensive score of teaching interaction efficiency corresponding to each training segment is calculated.

5. The method for evaluating flight training operation efficiency based on flight operation data according to claim 4, characterized in that, The specific process for calculating the trainee's comprehensive skill score in each training session of a specific training subject is as follows: Obtain historical training segments for each student in a specific training subject, extract key performance indicators (KPIs) from them, and select the KPIs corresponding to each training session. Based on a pre-defined standardization function, each key performance indicator is mapped to a corresponding interval. Combining the pre-defined standardization function and the weight coefficients of each key performance indicator for each training session, the comprehensive skill score of the trainee for each training session for a specific training subject is calculated.

6. The method for evaluating flight training operation efficiency based on flight operation data according to claim 5, characterized in that, The specific process for establishing a skill decay model for trainees in a specific training subject is as follows: Obtain the comprehensive skill score sequence of trainees in a specific training subject and the timestamp of each training session. Based on the comprehensive skill score sequence and the corresponding timestamp, fit a skill decay model of trainees in a specific training subject.

7. The method for evaluating flight training operation efficiency based on flight operation data according to claim 6, characterized in that, The specific process for predicting the student's skill gap at any future point in time is as follows: The system obtains the date on which the trainee plans to conduct the next training session. Based on the skill decay model, it calculates the trainee's predicted skill score for a specific training subject on that training date. Combined with a pre-set skill attainment threshold, it calculates the quantitative value of the trainee's predicted skill gap for a specific training subject on that training date. Based on the trainee's predicted skill gap quantification value for a specific training subject on the training date and the preset skill attainment threshold, assess whether the trainee has a skill gap for a specific training subject on the training date; If it exists, obtain the student's average skill score improvement value for a single training session for a specific training subject, and combine it with the predicted skill gap quantification value to calculate the student's additional training sessions for that specific training subject on that training date.

8. The method for evaluating flight training operation efficiency based on flight operation data according to claim 7, characterized in that, The specific process for calculating the rate of decline of observed skills between adjacent training sessions in a specific training subject is as follows: Obtain the trainee's comprehensive skill score corresponding to adjacent training segments for a specific training subject; Obtain the actual interval duration between adjacent training segments for a specific training subject for that student; The rate of decline of observational skills between adjacent training segments in a specific training subject was calculated.

9. The method for evaluating flight training operation efficiency based on flight operation data according to claim 8, characterized in that, The specific process for assessing the impact of the overall score on the efficiency of teaching interactions on the skill decay rate is as follows: A related dataset is constructed based on the comprehensive score of teaching interaction efficiency of each training segment and the observed skill decay rate between adjacent training segments in a specific training subject; Statistical analysis was performed on the associated datasets to obtain the parameters of the correlation model that quantifies the influence of the comprehensive score of teaching interaction efficiency on the rate of decline of observed skills.

10. The method for evaluating flight training operation efficiency based on flight operation data according to claim 9, characterized in that, The specific process for generating real-time feedback reports and flight training plans for trainees is as follows: Based on the comprehensive score of teaching interaction efficiency, the personal skill decay model and the correlation model, optimization strategies are generated, including real-time teaching feedback reports and flight training plans. Based on the comprehensive score of teaching interaction efficiency and its sub-indicators of the current training segment of the student, and by comparing with historical data, a real-time teaching feedback report is generated for the student. Based on the prediction results of the skill decay model, the parameters of the correlation model, and the preset skill attainment requirements, the recommended training intervals are calculated and the training subjects that need attention are identified, thereby generating the corresponding flight training plan for the trainees.

Images