Friction stir welding tool condition monitoring using vibration signals and Random forest algorithm-A Machine learning approach

Friction stir welding (FSW) is a new kind of solid-state welding technique. Rigid and reliable joints in intricate shapes are possible with FSW. This type of welding process is frequently used in many commercial applications like automobile, shipbuilding, aerospace, and many more. The monitoring of the welding tool condition is essential for the inline process to identify and avoid the early defect of the workpiece and the breakdown due to the increase of the ideal condition of the machinery. Condition monitoring of the FSW tool is an advanced and novel predictive maintenance technique, in which the real-time vibration data are collected from the FSW machine under different operating conditions using an accelerometer sensor. The acquired vibration signals were analyzed using the machine learning approaches through feature extraction and feature classification. This paper aims the vibration analysis based on FSW tool condition monitoring using machine learning algorithms such as decision tree, Logistic Model Tree (LMT), Hoeffeding, and Random forest. Among all classifiers, Random Forest gives better results. (c) 2021 Elsevier Ltd. All rights reserved. Selection and peer-review under responsibility of the scientific committee of the 28th International Conference on Processing and Fabrication of Advanced Materials.