| Popis: |
Most non-destructive tests of concrete require sophis-ticated equipment and training; in this work we aim to develop a simple method to estimate the strength class of cylindrical con-crete samples based on vibrations signals that are collected after striking a concrete cylinder with a hammer. The vibration signals were collected by attaching a smartphone to the concrete cylinder and logging the vibrations registered via the smartphone’s built-in accelerometer. The acquired 1-D vibration signals are trans-formed to 2-D scalograms using continuous wavelet transform. Scalograms are then used to train a deep learningmodel to predict the strength class. Preliminary findings show that the model is capable of classifying the strength of concrete to low, high, or me-dium. The developed model achieved a high accuracy of 91.67%. The promising results of this work shed light into the future of smartphone-based measurements of construction materials’ properties. |
