高铁震源地震信号的稀疏化建模

我国每天有数千趟高铁列车运行在两万多公里的高铁线路上，不但会引起高铁路基的振动，还会激发出地震波.地震检波器所接收到的数据中不仅包含窄带分立谱特性的高铁震源地震信号，还包含宽频带特性的背景信号.如何实现从检波器所接收到的高铁震源地震数据中分离出高铁震源地震信号和宽频带背景信号是准确利用该类信号的关键.考虑到高铁震源地震信号与宽频带信号在频率域明显的形态特征差异，本文首次将形态成分分析这种信号分离手段引入到高铁震源地震信号处理中，实现高铁震源地震信号的稀疏化建模并进而实现从接收数据中分离出高铁震源地震信号以及宽频带背景信号.对北京大学在中国南方某高铁沿线采集到的大量高铁震源地震数据进行处理，结果表明：采用形态成分分析并结合分块坐标松弛算法，能够实现实际采集高铁震源地震数据中的高铁震源地震信号和宽频带信号的分离.

Sparse modeling of seismic signals produced by high-speed trains

In China, there are thousands of high-speed trains running on over twenty thousand kilometers railways every day. These running high-speed trains not only cause vibration of railway roadbed but also stimulate seismic waves. The data recorded by seismic receivers to these events contain multi-narrow-band waves by high-speed trains as well as wide-band background noise. Therefore, the key step to use these signals is to divide the complex data into the real high-speed train induced seismic signals and the background noise. Considering the morphological difference between the high-speed train induced seismic signals and the background noise, we firstly introduce the morphological component analysis to the processing of the high-speed train induced seismic signals, and realize the separation between these signals and the background noise by sparse modeling. We applied this approach to the dataset collected by Peking University in South China. The results show that the acquired dataset can be successfully divided to the multi-narrow-band high-speed train induced seismic signals and the wide-band background noise by using the morphological component analysis and block-coordinate relaxation.