Condition Monitoring in Additive Manufacturing Using Support Vector Machine

Additive manufacturing (AM) involves the deposition of materials to form a three-dimensional object by printing. Presently fused deposition modelling (FDM) is one of the most widely used AM technique because of ease of operation and cheaper product. To get better part quality, there is a need to identify and monitor any process failure during 3-D printing. In this paper, the experimental data for the faulty and healthy condition of the printed specimen is collected using accelerometer at different process parameters. During time-domain feature selection root mean square (RMS), interquartile range (IQR) and mean absolute deviation (MAD) were identified as key features for classification. When root mean square and mean absolute deviation were used as the main features for training the FDM model based on a quadratic support vector machine algorithm (SVM) and a K-fold cross-validation approach, an accuracy of 78.6% is achieved. Such a technique is capable of preventing the faulty component in the job floor and help save the material by diagnosing the machine at the earliest.