Inverse PageRank-Based Fault Propagation Analysis of Production Lines under Multiple Uncertain Performance Indicators

In production lines, quality defects and safety incidents often originate from the propagation of faults resulting from the coupling of multiple factors. After fault propagation, multiple performance indicators on coupling devices are influenced, such as reduction of workpiece quantity, and decrease of processing time. However, the variation of indicators cannot definitively confirm the occurrence of fault propagation, it only provides uncertain information. This credibly uncertain information from various indicators can be conflict, leading to difficulties in analyzing fault propagation issues. To this end, an inverse PageRank-based method is developed for fault propagation analysis under multiple uncertain performance indicators. To fuse multiple uncertain indicators, the Dempster-Shafer evidence theory is adopted. Note that a piece of equipment is harmful if the failure generated from it can spread to other important equipment, which is in line with the idea of inverse PageRank. To assess the fault propagation risks, the inverse PageRank algorithm is used resorting to the fusion information. Subsequently, the fault propagation path can be identified by means of the obtained propagation risks. Finally, the method is validated on a real impeller production line.