Securing cyber-physical additive manufacturing systems by in-situ process authentication using streamline video analysis

Abstract

In cyber-physical systems (CPS) of additive manufacturing (AM), cyber-attacks may significantly alter the design of the AM part, compromising its mechanical properties and functionalities. In-situ process authentication may assure that the AM part is fabricated as intended. Most cyber-physical attacks towards AM processes can be manifested as printing path alterations, and an in-situ optical imaging system can detect alteration in the printing path. This will prevent catastrophic geometric changes and mechanical property compromises in the AM parts, ultimately improving the AM process security. In this paper, a novel process authentication methodology is proposed based on image texture analysis of the layer-wise in-situ videos. The layer-wise distribution of the segmented textures’ geometric features is characterized as the layer-wise texture descriptor tensor (LTDT). Given the high dimensionality and sparsity of the extracted LTDTs, the multilinear principal component analysis (MPCA) algorithm is used for dimension reduction. Subsequently, the Hotelling T2 control charting technique is adopted for alteration detection based on the extracted low-dimensional layer-wise features. Case studies based on a fused filament fabrication (FFF) process were conducted to evaluate and validate the proposed framework. The proposed method can achieve over 95% of accuracy, which illustrates that the proposed method can accurately detect process alterations due to printing path changes. In addition, the proposed method significantly outperforms the benchmark method. The computation time for both the proposed and benchmark method is also compared.