铁路桥梁动态监测数据压缩技术研究

发布时间：2018-05-27 11:08

本文选题：铁路桥梁 + 健康监测　；参考：《石家庄铁道大学》2015年硕士论文

【摘要】：随着运输业的蓬勃发展,桥梁在铁路线路中所占比重越来越大,发挥着越来越重要的作用,然而,随着桥梁数目、跨度、运营时间以及荷载的不断增加,其运营安全性逐渐引起人们的广泛关注,因此桥梁健康监测系统的产生成为一种必然趋势,监测系统的建立和安装使得桥梁管理者能够实时掌握桥梁的运营状况及安全状态,及时发现安全隐患,减少事故的发生。但是,由于监测系统的长时性,将会产生海量的数据,这会对数据的传输和存储带来不便,因此,为了解决结构健康监测当中通信带宽受限以及节约存储空间等问题,对于监测数据的压缩处理研究成为发展的必然趋势。本文针对铁路桥梁健康监测数据的海量性,考虑桥梁动荷载特性以及周围环境的影响,结合数据的周期性、循环平稳性等特点,利用压缩感知原理,提出针对铁路桥梁健康监测数据的分析和重构算法。由于铁路桥梁上运行列车编组的固定性以及车辆参数的一致性,使得监测数据呈周期性变化,本文首先通过ANSYS仿真验证其周期性的存在,再通过小波变换分析、循环谱分析等方法探索实测数据的内在周期性,为后续分析做好铺垫。压缩感知理论是不同于传统采样定理的一种新兴理论,在采样过程当中不在局限于带宽的限制,本文根据监测数据的特点,在深入分析压缩感知原理的基础上,对比各种重构算法的优缺点,采用OMP算法对监测数据进行压缩和重构,取得了较好的效果,证明了该算法的可行性。稀疏贝叶斯学习是近几年新兴的一个研究领域,将其与压缩感知理论相结合对于块状稀疏信号具有较好的重构效果,同时也克服了OMP算法计算复杂、迭代时间长的缺点。由于监测信号呈周期性变化,因此通过差分处理使其具有块状稀疏的特性,从而可以利用贝叶斯压缩感知进行压缩重构,同样取得了较好的重构效果;为了使压缩重构过程更加简单有效,本文采用基于稀疏贝叶斯学习的压缩感知算法对未经过处理的加速度信号(非稀疏性)进行压缩和重构,取得了较好的重构效果,证明了该算法的可行性和实用性。
[Abstract]:With the rapid development of transportation industry, bridges play a more and more important role in railway lines. However, with the increase of the number of bridges, span, operation time and load, The safety of the bridge has been paid more and more attention, so the generation of the bridge health monitoring system has become an inevitable trend. With the establishment and installation of the monitoring system, bridge managers can grasp the bridge operation status and safety status in real time. Discover safety hidden danger in time, reduce the occurrence of accident. However, because of the long-term nature of the monitoring system, a large amount of data will be produced, which will bring inconvenience to the transmission and storage of data. Therefore, in order to solve the problems of limited communication bandwidth and save storage space in structural health monitoring, The research on compression and processing of monitoring data has become an inevitable trend of development. Aiming at the magnanimity of railway bridge health monitoring data, considering the dynamic load characteristics of bridge and the influence of surrounding environment, combined with the characteristics of periodicity and cycle stability of the data, the principle of compression sensing is used in this paper. An algorithm for analysis and reconstruction of railway bridge health monitoring data is proposed. Because of the fixed train formation and the consistency of vehicle parameters on the railway bridge, the monitoring data change periodically. Firstly, the existence of periodicity is verified by ANSYS simulation, and then the wavelet transform is used to analyze it. Cyclic spectrum analysis and other methods are used to explore the periodicity of measured data so as to pave the way for subsequent analysis. Compression sensing theory is a new theory which is different from the traditional sampling theorem. It is not limited to bandwidth in the sampling process. According to the characteristics of monitoring data, this paper analyzes the theory of compressed sensing in depth. Compared with the advantages and disadvantages of various reconstruction algorithms, the OMP algorithm is used to compress and reconstruct the monitoring data, and good results are obtained, which proves the feasibility of the algorithm. Sparse Bayesian learning is a new research field in recent years. The combination of sparse Bayesian learning with compression sensing theory has a good reconstruction effect for block sparse signals, and it also overcomes the shortcomings of complex computation and long iteration time of OMP algorithm. Because the monitoring signal changes periodically, it has the characteristic of block sparseness by differential processing, so it can be compressed and reconstructed by Bayesian compression perception, and good reconstruction effect is also achieved. In order to make the process of compression and reconstruction more simple and effective, the compression sensing algorithm based on sparse Bayesian learning is used to compress and reconstruct the unprocessed acceleration signal (non-sparsity). The feasibility and practicability of the algorithm are proved.
【学位授予单位】：石家庄铁道大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U446

【相似文献】