Research And Implementation Of Data Processing Methods For Key Components Of EMU

With the development of China's high-speed railway and the arrival of the era of big data,data analysis is becoming more and more important in the healthy management of high-speed EMU.For comprehensive and accurate data analysis,it needs to obtain high-speed railway EMU data from multiple information systems.However,multi-source data has problems such as large amount of data,heterogeneous,multi-dimensional,multi-scale,asynchronous,and low data usage efficiency,which increases the difficulty of data analysis and even will make the analysis result the opposite of the actual situation.Therefore,it is especially important to unify and standardize the data of the high-speed EMU before analyzing these multi-source data.In view of the large amount of data,heterogeneous,multi-dimensional and multi-scale characteristics in the multi-source data.This paper proposes a targeted ETLTL(Extract-Transform-Load-Transform-Load)two-stage data cleaning process,improves standard data warehouse multi-tier architecture and traditional denoising method based on EMD.Based on this,a cloud-based big data management platform integrating data collection,storage and providing is established,which manages data distributed across different data sources in a unified,standardized,and efficient manner and provides a high-quality source of data for EMU data analysis.The main work of this thesis includes the following three points:(1)Through the analysis of the multi-source data of the EMU's whole life cycle,the characteristics of large data volume,heterogeneous,multi-dimensional and multi-scale are obtained.In order to meet the practical needs of multi-faceted data analysis under existing information system resources,we propose a targeted ETLTL two-stage data processing flow,and the standard data warehouse multi-layer architecture system is improved to meet the actual data management requirements.(2)We analyze the empirical mode decomposition(EMD)method and the traditional EMD-based denoising method.Also,we find and improve the problems in them.Then,we de-noise the different motor speeds and bearing vibration signals of different fault conditions by traditional method and improved method separately.By comparison,the experiment results perform that the improved method is more effective.(3)Last,based on the previous theoretical research results,data processing was implemented from three aspects of data collection,data storage and data provision.Finally,the data management cloud platform of the high-speed highway EMU life cycle was established,and the whole process of data processing was completed.