Research On The Method Of Improving The Quality Of Flight Training Data

In recent years,with the rapid development of the global civil aviation industry,China’s civil aviation education and training industry is also rapidly rising,the demand for pilots is increasing sharply,and the training tasks are increasing.At present,all flight training aircraft are equipped with flight data recorders,which are used to record various data generated during the flight.In the process of flight training,a large number of data resources will be generated.These data truly record the process of flight training of students,and contain various information such as students’ control skill level,safe driving behavior characteristics,flight safety status and so on.A full understanding and application of this information will help improve the quality of flight training and improve the level of flight safety.However,the acquisition of massive flight training data is delayed in time,and the multi-source and fuzzy characteristics make it difficult to directly use it for specific applications and research.Therefore,it is necessary to conduct research on flight training data,propose methods to collect data quickly,select appropriate algorithms,develop a system for improving the quality of flight training data,reduce low-level repetitive research work,improve the application rate of data,and better Lay the foundation for flight training and related research.The purpose of this thesis is to solve the problems of delayed acquisition of flight training data and insufficient comprehensive application,realize intelligent analysis of data,protect privacy,and improve the availability of data,so as to obtain highly available flight training data,and establish a standard process-based system.Data processing tools,in order to realize the transformation from unavailable raw data to high-quality flight training data,provide safe and reliable data for related applications,and lay the foundation for further research on flight training-related applications in the future.Based on the data recorded on the Cessna 172 R training aircraft of the Civil Aviation Institute of China,the research on flight training data is divided into three parts: the first part is data collection.It is tested;the second part is data preprocessing,using robust random cutting tree algorithm for outlier detection,KNN algorithm for missing value estimation;the third part is data quality improvement,using linear interpolation to improve accuracy,Kalman filtering algorithm Perform denoising and smoothing,offset time,desensitize the coordinate system of latitude and longitude conversion,and finally compare and verify the processed data with the original data;the fourth part is system development,integrating the above algorithms to complete flight training Construction of data quality improvement system.