Research And Realization Of Emu’s Operation And Maintenance Decision-making Recommended Techniques Based On Knowledge Base

With highly development and widely spread used, EMU has being played an important role in passenger and cargo traffic, which possesses efficient, safe, fast and high standards service. However, it is inevitable that failures would show up occasionally during the EMU operating services. So the priority task is to how to guarantee a efficient and safe operation of EMU and how to make full use of the collected data to establish operation and maintenance services system of EMU and further support the services, which has become a mainly studied area. Currently, the predominant pattern of EMU operation and maintenance is planned maintenance, which is likely to result in a waste of resources and low efficiency in failure maintenance. With the number of EMU increasing and the second generation WTD(Wireless Transmission Device) technology coming into use, it is more difficult for the existing system to adapt to the data processing and the speed of data retrieval is being slower and slower, as sharp increase data collected from EMU.In order to deal with the above issues, a new operation and maintenance decisions recommended system of EMU was designed and developed in this paper. The system can mainly estimate the real-time missing data, recommend potentials failure of online EMU by using recommendation technology and provide exact decision-making for different failures separately by solution knowledge base. Specifically, an improved Bayes algorithm was used to estimate the real-time missing data, which comes from key components of EMU during the work, taking the traction motor for an example to verify it. Then, make failure recommendation for on-line EMU by using collaborative filtering algorithms based on the real-time status data, and further provide accessible scheme for failure by solution knowledge base. Finally, these methods are applied under the Hadoop framework, solving the problem of magnanimity and inefficiency of data processing.Experiments show that the improved Bayes algorithm regards to completing the real-time data of key EMU components can improve the quality of data used for data mining. Using collaborative filtering recommendation algorithm for failure recommendation can facilitate to improve the efficiency of operation and maintenance and the speed of failure repair. The efficiency of operation and maintenance of EMU was increased by improving the speed of data processing through realizing these methods with MapReduce.