IEEE Project Abstract

Automated fault detection is an important part ofa quality control system. It has the potential to increase theoverall quality of monitored products and processes. The faultdetection of automotive instrument cluster systems in computerbased manufacturing assembly lines is currently limited tosimple boundary checking. The analysis of more complex nonlinear signals is performed manually by trained operators,whose knowledge is used to supervise quality checking andmanual detection of faults. We present a novel approach forautomated Fault Detection and Isolation (FDI) based on deeplearning. The approach was tested on data generated bycomputer-based manufacturing systems equipped with local andremote sensing devices. The results show that the approachmodels the different spatial/temporal patterns found in the data.The approach can successfully diagnose and locate multipleclasses of faults under real-time working conditions. Theproposed method is shown to outperform other established FDImethods.