| This thesis aims to build a big data resource pool for flight training to solve the problems of scattered training flight data and difficult data access.As a basis for detecting flight anomalies and improving flight quality,flight parameter data has an extremely important role and value.increase.In response to this demand,this thesis uses the flight parameter data collected by the sensor as the storage object,and builds a flight training big data resource pool based on the Hadoop technology framework to store the original flight parameter data and other related data that can be used for training flight research.Form a multi-source flight training raw data resource base;conduct further research on time series segmentation and semantic understanding of the preprocessed high-availability data,including automatic division of flight stages,automatic recognition of flight actions,and flight subject recognition,and the recognition results are stored in the resource pool,as the understanding and analysis of raw data,which can be used to support other applied research.The main research ideas of this thesis are as follows: firstly,acquire and store flight parameter data and flight-related data,then use the preprocessed data for semantic understanding to form sub-sequences of each flight stage and each flight subject;finally,divide the flight parameter data according to the flight stage and flight subject classification are stored for subsequent researchers to recall.Experiments show that the flight training big data resource pool constructed in this thesis can efficiently classify and store flight parameter data and training flight-related data,provide data resource support for flight training and related research in the field of big data,and identify methods for flight subjects.It also has better practicability and accuracy. |
