Let’s Vibrate with Vibration: Augmenting Structural Engineering with Low-Cost Vibration Sensing

Using low-cost piezoelectric sensors to sense real structural vibration exhibits great potential in augmenting structural engineering, which is yet to be explored in the literature to the best of our knowledge. An example of such unexplored augmentation includes classifying diverse structures (such as buildings, flyovers, foot over-bridge, etc.). To explore these aspects, we develop a low-cost piezoelectric sensor-based vibration sensing system aiming to collect real vibration data from diversified civil structures remotely. We dig into our collected sensed data to classify five different types of structures through rigorous statistical and machine learning-based analyses. Furthermore, we design a lightweight Convolutional Neural Network architecture and perform necessary hyperparameter tuning to achieve better accuracy in classification. Our analyses achieve a classification accuracy of up to 97% with an F1 score of 0.97.