An Improvement of Physics Based Compressive Sensing With Domain Decomposition to Monitor Temperature in Fused Filament Fabrication Process
Abstract Sensors in manufacturing play an important role in monitoring and improving the quality of products. However, the rising cost of sensing subsystems and the bandwidth limitation of data transmission are challenges in modern manufacturing systems, which rely on a large number of sensors. Recently, a physics based compressive sensing (PBCS) approach was proposed to monitor manufacturing processes with reduced number of sensors and amount of collected data. PBCS significantly improves the compression ratio from classical compressed sensing by incorporating the knowledge of physical phenomena in specific applications. In this paper, a modified orthogonal matching pursuit (OMP) recovery algorithm, called constrained OMP, is developed for PBCS when coherence exists between the measurement matrix and basis matrix. The efficiency of PBCS recovery is also improved by introducing a domain decomposition approach, which can reduce the size of model matrices, such as the conduction matrix and mass matrix in the transient heat transfer application. The improved PBCS with the domain decomposition method is used to monitor the temperature distribution in the cooling process and real-time printing process of fused filament fabrication.