Research On General Aviation Flight Training Quality Monitoring And Evaluation Model

With the continuous development of the aviation industry,it has become an important consensus for all civil aviation practitioners to vigorously train high-quality and high-level aviation personnel,construct a high-quality teaching system and strive to ensure the absolute safety of civil aviation.Taking the flight training industry as an entry point,this thesis explores the construction of a machine learning-based flight training over-limit event recognition model and an over-limit event-based flight training quality assessment model,aiming to help flight training institutions better identify over-limit events and discover training safety hazards and improve the quality of training and teaching.In this thesis,through the application research of flight training quality monitoring,the over-limit event monitoring items are prioritized.Airborne SD card data features,using python language to fill in missing values and data set division of sample data,and realize intelligent identification of flight training overrun events based on GBDT classification algorithm.The ACC curve,confusion matrix are used to evaluate the recognition effect of the model.According to the calculation results,the ACC value reaches more than 96%,and the performance indicators of the model are all at an excellent level;compared with the flight training evaluation index system,The CRITIC-G1 method based on the principle of variance maximization is used to optimize the combination of subjective and objective weights of the indicators,and the weighted TOPSIS model is used to achieve quantitative evaluation of flight training quality.At the same time,the RSR method is introduced to establish a grading standard to achieve the classification of flight training quality levels..By collecting the evaluation index data of an aviation school for 14 consecutive months for example analysis and verification,the model has a good evaluation effect,and the evaluation results are consistent with the actual situation of the training operation.The over-limit event identification model studied in this thesis overcomes the defects of the traditional "threshold + regular expression" identification method,and improves the efficiency of over-limit event identification;the constructed flight training quality evaluation model based on over-limit events not only realizes the Quantitative evaluation also achieves qualitative evaluation,which can provide decision support for training operation management.