Real-Time Fault Detection and Condition Monitoring for Industrial Autonomous Vehicles

Abstract

Real-time status monitoring and early defect diagnosis are becoming more and more necessary for modern industrial systems. Developing intelligent remote diagnostic technologies and integrating on-board and off-board diagnosis are two more unresolved research projects in the automotive industry. The automated transfer vehicle (ATV) equipment condition monitoring example in this chapter is part of a smart industrial use case. The suggested method successfully permits ATV fault scenarios to be seen in real time by expanding to a fleet of devices in an actual production plant. The application of a statistical threshold is the initial stage in creating a high-performance detection model for defect detection on stacked long short-term memory networks. For better computation speed, a second model trimming approach based on principal component analysis is suggested. The improved fault detection technique is ultimately applied by the airborne embedded computer platform with field-programmable gate arrays.