Data Driven Condition Monitoring Based on a Digital Twin for a Linear Actuator Realized As a Closed Hydraulic System

Linear actuators, implemented as closed hydraulic systems, without external piping, are a state of the art drive concept, see (Gannon, 2017). Collecting data, used to train condition monitoring (CM) for such drives, running 24/7, is cumbersome or even not possible. To gain training containing valid and invalid system states, we developed simulation model, consisting most relevant physical effects. The simulated data evaluated by one-step feature approach additionally with two-step using two less complex fault separation methods. In end, method showed be slightly better. only recognize, but also distinguish between accumulator pump faults.