EGU2020-1484
https://doi.org/10.5194/egusphere-egu2020-1484
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

TraML: separation of seismically-induced ground-motion signals with Autoencoder architecture

Artemii Novoselov, Gerrit Hein, Goetz Bokelmann, and Florian Fuchs
Artemii Novoselov et al.
  • University of Vienna, Department of Meteorology and Geophyscis, Vienna, Austria (artemii.novoselov@univie.ac.at)

Any time series can be represented as a sum of sine waves with the help of the Fourier transform. But such a transformation doesn’t answer whether the signal is coming from one source or several; neither it allows separation of such sources. In this work, we present a technique from the Machine Learning domain, called Auto-encoders that utilizes the ability of the neural network to generate signals from the latent space, which in turn allows us to identify signals from an arbitrary number of sources and can generate them as separate waveforms without any loss. We took ground motion records of passing trains and trams in the vicinity of the University of Vienna and trained the network to produce “clean” individual signals from “mixed” waveforms. This work proves the concept and steers the direction for further research of earthquake-induced source separation. It also benefits interference seismometry, since “noise” used for such research can be separated from the signal, thus reducing manual processing (cutting and clipping signals) of seismic records. 

How to cite: Novoselov, A., Hein, G., Bokelmann, G., and Fuchs, F.: TraML: separation of seismically-induced ground-motion signals with Autoencoder architecture, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-1484, https://doi.org/10.5194/egusphere-egu2020-1484, 2019

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 1 – uploaded on 28 Apr 2020
  • CC1: Comment on EGU2020-1484, Jonathan Bedford, 05 May 2020

    Dear Authors,

    Thank you for sharing your contribution. I enjoyed reading this. 

    I have a few questions:
    What is the input to the neural network?  Do you feed in a full time series?  Or just several samples of the time series?  Do you have to specify the number of outputs that you want (e.g.. user specifies that the signal is made of 3 sources) or is there a way around finding the number of signals that are mixed together?  

    Kind regards,

    Jonathan Bedford.

    • AC1: Reply to CC1, Artemii Novoselov, 05 May 2020

      Dear Authors,

      Thank you for sharing your contribution. I enjoyed reading this. 

      I have a few questions:
        

      Kind regards,

      Jonathan Bedford.

      Hello, dear Jonathan!
      What is the input to the neural network? - We feed a mixture of two time-series (signal1 + signal2) in numpy array format. The length of this signal is 4000 samples, which corresponds to 40 seconds (note that a typical train signal recorded by our station is ~30 sec) .

      Do you have to specify the number of outputs that you want (e.g.. user specifies that the signal is made of 3 sources) or is there a way around finding the number of signals that are mixed together?  
      At the moment, we do specify the number of sources, but we are working on automatic source counting as we speak now, with the hope that this will increase the model's ability to generalize