Research And Implementation Of Key Technologies In The Analysis Of The Relationship Between Faults In Large Data Sets Of EMU

With the high-speed development of the EMU and the use of large-scale, EMU has become the important tool for railway passenger and cargo,which with efficient, safe, rapid, high standards of service function, plays an important role in traffic is also easy to highlight. High-speed railway in China after ten years of operation, has accumulated a lot of failure data, and every day after receipt of the new data, how to dig out the effective information and knowledge,for the EMU running safety, faultdiagnosis and maintenance decision to provide support,is of great significance.In recent years, high-speed railway technology rapid development, the data accumulated presents exponential growth trend, the traditional association rule mining algorithm is difficult to meet the demand. Based on this, in this paper, the Eclat algorithm combined with the current epidemic MapReduce cloud computing platform, the mass of the EMU failure data association rules mining task decomposition in the cluster on multiple computers parallel processing, improve the efficiency of association rule mining.Usually database has two kinds of forms,include level data representation and vertical data representation, the common association rules algorithm such as Apriori algorithm and FP-Growth algorithm, they are based on the level of data representation. Eclat algorithm is one of the first association mining algorithm which using vertical data representation. Large Eclat algorithm in itemsets, intersection operation consumes a lot of time and system memory.First,the paper makes a deep research and analysis on the Eclat algorithm,aiming at the shortcomings of the Eclat algorithm and the understanding of MapReduce, suggestedan improved algorithm based on MapReduce, and applied the impoved algorithm to the EMU failure Association Rule Mining. Compared with the traditional mining algorithms, the impoved algorithm have significantly increased efficiency.This paper finally also aims at the malfunctions of the EMU hierarchies, the importance of fault data inconsistency suggested a weighted association rule mining algorithm, excluding the meaningless association rules mining, priority mined rules that contain high fault level, in line with the practical requirements,to the emu fault detection and prevention, fast fault location to provide help.