The Building Data Warehouse Based On QAR And Application To Fault Analysis

Data warehouse is one technology of data storage and management for online analysis and data mining. It makes the original operational data of different sources processed and integrated, and provides complete, timely, accurate and intuitive decision support information for enterprise. Data warehouse is built based on comprehensive and perfect information applications, and used to do analysis of high decision support.Firstly, the related current research status at house abroad is introduced of the aircraft fault diagnosis and estimation, and data warehouse. By the study on formats of QAR(Quick Access Recorder) data and characteristics of data warehouse, a method is put forward,which uses the technology of data warehouse to analyze and predict the air accidents.Secondly, the established solution model of QAR data warehouse is introduced,and the design step and realization method is mainly presented. With the subject of flight safety, the relevant QAR parameters are defined as well. During the design, according to the feature that QAR data volume is high, the multiple storage strategy is adopted to achieve a balance between the performance and storage space of data warehouse. And during the realization,the method which based on the tool of ETL(Extract, Transform, Load) and SQL is adopted to realize the ETL process, for the development speed and efficiency of ETL is clearly enhanced.Finally, based on QAR data warehouse, fault diagnosis is analyzed by OLAP and data mining. The multi-dimensional cube is established, by which slice query is made from different aspects, and the visual OLAP analysis is also realized. Then, owing to the time-serial feature of QAR data, the developing trend of aircraft fault is predicted with time-serial algorithm. Because of the uncertainty about fault, the faults with uncertainties are analyzed by using naive bayesian algorithm. Then using association rules algorithm, the association rules are discovered. The experiment results show that, the prediction of trend and association analysis can provide effective decision-making support information in developing the professional skills of crew, strengthening the control of the safety, ensuring the good development in aviation and so on.