Detecting Malicious Defects in 3D Printing Process Using Machine Learning and Image Classification

Abstract

3D printing, or additive manufacturing, is a key technology for future manufacturing systems. However, 3D printing systems have unique vulnerabilities presented by the ability to affect the infill without affecting the exterior. In order to detect malicious infill defects in 3D printing process, this paper proposes the following: 1) investigate malicious defects in the 3D printing process, 2) extract features based on simulated 3D printing process images, and 3) an experiment of image classification with one group of non-defect infill image and the other group of defect infill training image from 3D printing process. The images are captured layer by layer from the top view of software simulation preview. The data extracted from images is input to two machine learning algorithms, Naive Bayes Classifier and J48 Decision Trees. The result shows Naive Bayes Classifier has an accuracy of 85.26% and J48 Decision Trees has an accuracy of 95.51% for classification.

Detecting Attacks in CyberManufacturing Systems: Additive Manufacturing Example

“…This work is based on Wu et al [3] but extends it with i) additional machine learning methods, ii) more types of infills and iii) use of a camera in collecting and processing real images from 3D printing machines, and iv) application of machine learning to detect malicious defect in real images. The steps of installation of a camera, image processing and detection results shows the potential for implementing this 3D printing vision-based detection system in a real CyberManufacturing Systems.…”

“…This work builds on the previous research [3], but extends further by: 1) using real images collected by an Arducam OV2640 2 Megapixels Camera. Relevant changes are made to address the differences between the simulated images and real images.…”

“…A recent study by IBM reports that manufacturing is the second most frequently targeted industry in 2015 in terms of the number of cyber-attacks [2]. The consequence of cyber-attack on manufacturing systems could be defected parts, the cost of poor quality, damaged equipment, and loss of customer loyalty or even create customer safety issues [3].…”

“…The previous research by Wu et al [3] defined five different infill defect patterns: Seam, Irregular Polygon, Circle, Rectangle, and Triangle. The research used Naive Bayes Classifier and J48 Decision Trees classification algorithms to differentiate the defective and nondefective parts.…”

Automated Real‐Time Detection and Prediction of Interlayer Imperfections in Additive Manufacturing Processes Using Artificial Intelligence

“…It turns out that these imperfections are largely dictated by the settings of printing parameters, first‐layer calibration, and model geometry . With recent advances in applying artificial intelligence and machine learning to materials science and engineering problems, researchers have started using machine‐learning algorithms to classify and predict different printing defections including blob, warp, and delamination based on the settings of printing parameters . Another interesting approach in the field adopts a new slicing mechanism which splits prints into spatially locked bricks to reduce warping .…”