Machine Learning Assisted Metal Corrosion Detection Based on Back-Scattering Sensor

Here a metal corrosion detection system based on back-scattering sensor is proposed, with aid of machine learning classification algorithms. The data are obtained through full-wave simulation, then multi-label modeling and classification are carried out. The effects of different classification algorithms are compared. Among them, the random forest classifier has the best performance and the classification accuracy rate is up to 94%.