SeismoSense

Inspiration

For the Seismology challenge, we were inspired by the shared frequency band of frequency that the three types of target noise exhibited.

What it does

Finds certain types of seismic noise, characterized by a modulating frequency (sinusoidal, rising or falling edge).

How we built it

Applied bandpass filter to time-domain data and then took short time Fourier transform to see how frequency changes over time. The majority of the project focused on distinguishing the primary frequency of interest and transforming the 3D amplitude vs frequency and time graph into a frequency vs time graph that resembled the "visible" spectrogram lines as accurately as possible. We then took the Fourier transform of of frequency in time and characterized the noise based on the ratio of frequencies within the desired noise bracket. The interval over which we applied this shifted in each iteration to produce time stamps for each type of desired noise and it's frequency in time modulation frequency.

Challenges we ran into

With only two team members and little to no coding experience, we had trouble getting started and downloading the necessary software. We also struggled distinguishing clean functions from spectrogram lines, due to our unfamiliarity with computer vision.

Accomplishments that we're proud of

We are proud of the data analysis techniques we thought of at this hackathon.

What we learned

We learned to think more deeply about the amazing capabilities of human vision, how to effectively manage and visualize data, and how different analysis methods respond to certain properties of the data.

What's next for SeismoSense

The next step for SeismoSense would be comprehensive testing of its accuracy and a user interface.

Built With

• matlab