Condition monitoring of robot joints using statistical and nonlinear dynamics tools

I. Trendafilova, H.H. Van Brussel

Research output: Contribution to journalArticlepeer-review

5 Citations (Scopus)


This paper considers the problem for condition monitoring of robot joints employing measured acceleration signals. The study aims at (1) Determining features, to be extracted directly from the measured acceleration signals, to detect defects in robot joints and at (2) Finding features dependent on the size of the fault in order to quantify the present defects. The signals coming from intact robot joints and from joints containing backlash or clearance are analyzed using nonlinear dynamics as well as statistical tools. A method for defect detection that employs nonlinear autoregressive (AR) modeling of the acceleration signals is successfully applied to detect backlash and clearance in robot joints. Two procedures for defect quantification are considered - one of them based on the AR modeling and the other employing nonlinear dynamics and statistical features. The problems are considered in the context of a pattern recognition paradigm.
Original languageEnglish
Pages (from-to)283-295
Number of pages12
Issue number2
Publication statusPublished - 2003


  • condition monitoring
  • nonlinear dynamics
  • pattern recognition
  • signal processing
  • mechanical engineering


Dive into the research topics of 'Condition monitoring of robot joints using statistical and nonlinear dynamics tools'. Together they form a unique fingerprint.

Cite this