Abstract
Due to its predominant flexibility in fabricating complex geometries, additive manufacturing (AM) has gain increasing popularity in various mission critical applications, such as aerospace, health care, military, and transportation. The layerby-layer manner of AM fabrication significantly expands the vulnerability space of AM cyber-physical systems, leading to potentially altered AM parts with compromised mechanical properties and functionalities. Moreover, internal alterations of the build are very difficult to detect based on traditional geometric dimensioning and tolerancing (GD&T) features. Therefore, how to achieve effective monitoring and attack detection is a very important problem for broader adoption of AM technology. To address this issue, this paper proposes to utilize side channels for process authentication. An online feature extraction approach is developed based on autoencoder to detect unintended process/product alterations caused by cyber-physical attacks. Both supervised and unsupervised monitoring schemes are implemented based on the extracted features. To validate the effectiveness of the proposed method, two real-world case studies are conducted on a fused filament fabrication (FFF) platform equipped with two accelerometers for process monitoring. Two different types of attacks are implemented. The results demonstrate that the proposed method outperforms conventional process monitoring methods, and can effectively detect part geometry and layer thickness alterations in real time.
Cyber—Physical Attack Detection in Water Distribution Systems with Temporal Graph Convolutional Neural Networks
