Research On Detection Algorithm Based On Rough Set For QAR Outlier Data

Current domestic airlines generally use QAR(Quick Access Recorder, QAR) for the aircraft flight data records. QAR records every moment of the true state of the flight,The data was strongly basis of the real-time status monitoring of aircraft and late diagnosis.Particularly, the rapid growth of aircraft maintenance workload, limited troubleshooting time and Troubleshooting and maintenance efficiency is very low. Under these conditions, the QAR data for in-depth research to improve the aircraft's level of fault diagnosis is particularly important.In this paper, the characteristics of QAR flight data and rough set theory has been focused on the study. combination of these two factors together,by the knowledge of decision information table which was base of rough set theory, Detection of abnormal data in QAR data, obtained the important decision-making rules form it. Ultimately, through these rules were provided important decision support to the aircraft fault diagnosis, greatly improving the efficiency of aircraft troubleshooting.First, the analysis of the characteristics of QAR data, rough set theory and multi-attribute decision-making and then Based on rough set theory, QAR data mining process was established, Specifically, QAR data reading, data pre-processing, QAR data discreting, timing information system to convert non-timing information systems, the decision table attribute reduction and final rules extraction, and so on. From theory to practical application methods, every step of the process had done a detailed description and analysis of research. And the corresponding step discretization, attribute reduction and rule extraction is given the improved algorithms or new processing algorithms. Eventually,through the establishment of QAR data failure information decision table experiments were carried out, through the experiment, the data given by experts and related items prove its feasibility and effectiveness, thus, to make it more realistic with the meaning of Improveing the efficiency of aircraft troubleshooting.