Observer-based fault diagnosis in manipulation robots

The problem of fault detection and isolation (FDI) in manipulation robots described by nonlinear dynamic models is studied. It is supposed that the FDI process is based on the bank of diagnostic observers. To design the nonlinear observers, original mathematical technique (so-called algebra of functions) is used. The procedures for obtaining the diagnostic observers based on the algebra of functions are developed. Different kinds of sensors and their influence on a quality of the FDI process are studied. The modeling results reflecting the relationships between the faults and the residuals are presented. Use of the FDI methods allows us to prevent the destruction of equipment and avoid great losses. (C) 1999 Elsevier Science Ltd. All rights reserved.