高速列车海量数据故障分析系统研究

Study on System for Fault Diagnosis of Massive Data of High-speed Train

高速列车速度快,传感器采集的频率高,数据更新速度快,因此采集到的数据量大,对故障诊断系统的数据存储、管理和分析能力提出了巨大的挑战。针对目前的高速列车故障诊断系统难以有效处理海量数据的情况,设计了一种基于大数据的高速列车海量数据故障分析系统。采用当前流行的大数据处理工具构建了能够处理海量高铁数据的大数据基础支撑平台,能够高效的存储、管理高铁的海量传感器数据。并且针对传统故障诊断方法在海量数据的时候建模速度很慢甚至无法建模求解的问题,在大数据基础支撑平台的基础上研究了能够对海量高铁数据进行建模的分布式算法,提高了海量高铁数据故障建模的效率。最后通过Speedup、Sizeup、Scaleup这3个性能指标对系统集成的PCA故障建模算法(以分布式PCA故障建模算法为例)的效率进行验证,验证结果表明系统中的分布式算法在3个性能指标上都有良好的表现,能够有效进行大数据量的故障分析,为高速列车的故障诊断提供可靠的保证。

Because of the high speed of high-speed train, the high frequency of sensor acquisition and data updating, the amount of data collected is large, which poses a great challenge to the abilities of data storage, management and analysis of fault diagnosis system. In view of the current high-speed train fault diagnosis system is difficult to deal with massive data effectively, a fault analysis system based on big data for massive data of high-speed train is designed. The system adopts the current popular Hadoop and Spark framework and technology for processing big data, which can efficiently store and manage the massive sensor data of high-speed train. And aiming at the fact that the speed of traditional fault diagnosis methods is slow and even unable to solve the problem of big data modeling. On the basis of above big data analysis system, distributed algorithms which can handle big data is studied, which solves the problem of fault analysis of massive data of high-speed railway. Finally, the efficiency of the distributed algorithm of system integration(taking distributed Bayesian algorithm as an example) is verified by three performance metrics of Speedup, Sizeup and Scaleup. The result shows that the proposed distributed algorithm in the system has good performance on the three performance indexes, and the fault analysis of large amount of data can be carried out effectively, which can provide reliable guarantee for fault diagnosis of high-speed train.