基于云计算的海量高铁噪声数据并行处理方法研究

发布时间：2018-01-05 09:15

本文关键词：基于云计算的海量高铁噪声数据并行处理方法研究　出处：《西南交通大学》2013年硕士论文　论文类型：学位论文

【摘要】：随着高速铁路的飞速发展,高速铁路的安全与舒适成为当前研究的一个热点问题。安装在列车上的传感器采集的噪声数据反映了列车的运行状况,并与列车的安全息息相关。然而在噪声数据采集的过程中由于种种因素的影响,采集的列车噪声数据中含有不同频率和特点的干扰数据,干扰数据直接影响了数据的分析与处理。研究表明预处理和滤波处理可以有效地去除数据中的干扰数据。然而,随着采集的数据量越来越大,而传统的预处理与滤波方法均采用的是单机处理的方式,效率低下,无法满足实际需求。云计算技术是解决上述难题的一项关键技术,其中的MapReduce模型可用于大规模数据的并行运算,由于其良好的并行效果且不用了解其底层架构,目前已有很多学者利用MapReduce进行算法设计,且取得了良好的成果。因此本文拟将云计算技术应用到预处理与滤波方法中以提高列车噪声数据处理的效率,具有重要的实际应用价值。本文首先对预处理、滤波和云计算的国内外研究现状进行介绍。然后概述云计算技术与预处理方法,研究了预处理方法的并行化,提出了基于MapReduce的海量高铁噪声数据并行预处理算法,用Speedup和Sizeup并行化指标来评价算法的性能,实验结果表明并行预处理算法性能提升显著。紧接着,讨论了高通滤波、低通滤波、动窗滤波和中位值滤波等传统滤波技术,并对滤波方法进行并行化改进,提出基于MapReduce的海量高铁噪声数据并行滤波算法。波形展示和滤波正确性实验分析表明滤波效果明显。信噪比和均方差实验给出了高通滤波和低通滤波的最佳滤波参数。采用Speedup、 Sizeup和Scaleup这三个并行化参数评价并行滤波算法的性能,结果表明本文所提出的并行动窗滤波和并行中位值滤波算法性能提升显著；并行高通滤波和并行低通滤波算法由于使用了公共变量和受算法自身时间复杂度影响,并行效果受到一定影响。
[Abstract]:With the rapid development of high-speed railway, high-speed railway safety and comfort has become a hot issue in current research. The noise data collected by sensors installed on the train to reflect the running status of the trains, and is closely related to the safety of the train. However, due to various factors in the process of noise data acquisition, the data of different frequency interference and the characteristics of the noise data collected in the train with the interference data directly affects the data processing and analysis. The results indicate that the preprocessing and filtering can effectively remove the interference data. However, with the increasing of the data, while pretreatment with the traditional filtering methods are used is single the way, efficiency is low, can not meet the actual demand. Cloud computing is a key technology to solve the above problems, the MapReduce model can be used for large scale Parallel computing data, due to its good parallel efficiency and do not know the underlying architecture, many researchers have applied MapReduce to design algorithm, and achieved good results. So this thesis intends to apply cloud computing technology to pretreatment and filtering to improve the noise data processing efficiency, has important practical application value.
Firstly, preprocessing, research actuality of filtering and cloud computing. Then an overview of cloud computing technology and pretreatment methods, the study of parallel preprocessing methods, proposes a parallel preprocessing algorithm for high-speed railway noise data based on MapReduce, to evaluate the performance of the algorithm by using Speedup and Sizeup parallel index and the experiment results show that the parallel preprocessing algorithm performance significantly. Then, the high pass filter, low-pass filter, dynamic filtering technology of traditional filtering median filtering and window, and the filtering method is improved by parallel computing, parallel filtering algorithm proposed high-speed railway noise data based on MapReduce waveform display and filtering the correct. Experimental analysis shows that the filtering effect is obvious. The signal-to-noise ratio and the mean variance experiment provides the optimum filtering parameters of high pass filter and low-pass filter. By using Speedup, Sizeup and Scaleup three The performance evaluation parameters of parallel parallel filtering algorithm, the results show that the proposed median filtering algorithm and performance improvement action window filtering and parallel in parallel; parallel high pass filter and low pass filtering algorithm due to the use of public variables and time complexity of the algorithm itself, parallel effect affected.

【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TN911.4;TP338.6

【参考文献】