| 本文已被：浏览 215次 下载 251次 |
码上扫一扫！ |
| 基于分数阶理论的铁路基础设施形变监测数据分析与挖掘 |
| 刘亿,李平,封博卿,蒋丽丽,李聪旭,王虎,杨美皓 |
|
|
| (中国铁道科学研究院研究生部, 北京 100081;综合交通大数据应用技术国家工程实验室, 北京 100081;中国铁道科学研究院集团有限公司, 北京 100081;中国铁道科学研究院集团有限公司电子计算技术研究所, 北京 100081) |
|
| 摘要: |
| 北斗高精度定位系统因噪声等因素会产生一定范围内的随机误差，这导致传统模型难以直接对监测数据进行精准分析，因此提出基于分数阶理论的北斗监测数据分析方法，从数据整体趋势角度挖掘铁路基础设施形变演化规律。首先，给出分数阶分析方法的理论框架，并对基础理论进行详细介绍；其次，提出利用α稳定分布对原始数据进行概率密度拟合，实现数据的非高斯特性估计；再次，通过长程相关特性和多重分形特性挖掘隐藏在监测数据中的深层次趋势特征，从数据分数阶特性维度分析未来变化趋势；最后，所提分析方法应用于国内某重载铁路的基础设施形变监测，实验结果表明所提出的分析方法能够在噪声干扰下实现北斗监测数据的精准分析，能够对各组监测数据的演化规则和铁路基础设施形变程度进行精准判别。 |
| 关键词: 形变趋势分析 分数阶理论 铁路基础设施 长程相关特性 分形特性 |
| DOI： |
|
| 基金项目:中国国家铁路集团有限公司科技研究开发计划（N2021X033） |
|
| Analysis and mining of railway infrastructure deformation monitoring data based on fractional order theory |
| LIU Yi,LI Ping,FENG Boqing,JIANG Lili,LI Congxu,WANG Hu,YANG Meihao |
| (Postgraduate Department, China Academy of Railway Sciences, Beijing 100081, China; National Engineering Laboratory of Comprehensive Transportation Big Data Application Technology, Beijing 100081, China;China Academy of Railway Sciences Corporation Limited, Beijing 100081, China; National Engineering Laboratory of Comprehensive Transportation Big Data Application Technology, Beijing 100081, China;Institute of Electronic Computing Technology, China Academy of Railway Sciences Corporation Limited, Beijing 100081, China) |
| Abstract: |
| The BeiDou high-precision positioning system generates random error within a certain range due to the noise and other reasons, which makes it difficult to analyze the monitoring data directly and accurately for the traditional model. Therefore, the paper proposes a method of analyzing BeiDou monitoring data based on fractional order theory to mine the evolution line of railway infrastructure deformation from the perspective of data trend. Firstly, the theoretical framework of the fractional order analysis method is given and the basic theory is introduced in detail; Secondly, the raw data probability density is fitted by α-stable distribution and futher realizes the estimation of non-Gaussian characteristics; Thirdly, the deep-level trend features hidden in the monitoring data are mined through the long-range correlation characteristics and multifractal characteristics, and the future trend is analyzed from the fractional order characteristics dimension. Finally, the proposed analysis method is applied to the infrastructure deformation monitoring of a heavy hual railway line in China, and the experimental results show that the proposed analysis method can achieve accurate analysis of Beidou monitoring data under the interference of noise and can accurately discriminate the evolution rules of each group of monitoring data and the degree of railway infrastructure deformation. |
| Key words: Deformation trend analysis Fractional order theory Railway infrastructure Long-range correlation characteristics Fractal characteristics |
