Degradation Modeling of a Robot Arm to Support Prognostics and Health Management

Abstract

Abstract Robots are increasingly being adopted in manufacturing industries and this trend is projected to continue. However, robots, like all equipment, degrade once in operation and eventually fail. Yet today’s manufacturing systems are highly paced requiring high equipment availability. Tools and methods are being developed for monitoring, diagnostics, and prognostics to support maintenance activities. These tools require the presence of data representing both healthy and unhealthy states of the robot. Robot unhealthy state data is usually not available because robots are normally operated in a healthy state. A digital twin, which is a virtual real-time representation of a system, can support generating this data. This paper demonstrates the building of a digital twin of a robot workcell that inputs data from the real system. The most frequent robot degradations are identified as increased bearing friction and gear backlash, which are modeled in the digital twin. The digital twin is then used to generate data representing degraded states of the workcell, which are plotted against healthy state data to reveal patterns associated with the respective types of failure. The results show that modeling degradations in the digital twin can provide data which, when analyzed, can support prognostics and health management.