Machine Learning Based Surface Hardnes and Vibration Strength Estimation

Abstract

Using machine learning a vibration based system was produced that can infer different surfaces from one another. The work includes supervised learning with captured samples of an accelerometer. Using testing samples never used in training, system was tested to identify 10 different surface types. Main concept includes fast, non-destructive and blackbox style architecture to perform detection. All relevant software, libraries, hardware and developed technique included.