Early Malfunction Detection in Production Systems

The interaction of coupled subsystems may lead to a complex dynamics of the overall system. If failures arise it must be determined whether the failure can be allocated to a certain subsystem or if it is caused by the complex interaction of the individual subsystems.

In the framework of a project funded by the Federal Ministry for Education and Research (BMBF) and in co-operation with the enterprise ZF Friedrichshafen AG a software procedure has been developed, in order to detect, to classify and to locate malfunctions of a gear test facility. Self-generating neuronal networks, invented and developed by STASA, non linear modelling procedures and phase space reconstruction methods are used.