Spectral Network Approach For Multi-Channel Profile Data Analysis With Applications In Advanced Manufacturing

In the advanced manufacturing, a lot of sensors are used to collect real-time process signals for statistical monitoring. Motivated by the complex correlation structures of these mutli-channel profile signals, this paper proposes a monitoring scheme for their cross-correlations with the help of spectral network approaches. In particular, we first construct a network model for multi-channel profiles by extracting their features based on the multi-channel functional peA. The topological structure of the network can represent the cross-correlations of multi-channel profiles. Then we propose to monitor the topological structure using its spectrum information. Numerical studies in a certain fabrication process demonstrate the applicability and efficiency of the proposed methodology.