Development of data-driven PHM solutions for robot hemming in automotive production lines

Luca Actis Grosso, Andrea De Martin, Giovanni Jacazio, and Massimo Sorli
Publication Target: 
IJPHM
Publication Issue: 
1
Submission Type: 
Full Paper
AttachmentSizeTimestamp
ijphm_20_001.pdf1.78 MBFebruary 27, 2020 - 2:17am

Robotic roller hemming is currently one of the most used solution for joining metal sheets in automotive industry, especially for those production lines which need to favor flexibility with respect to raw productivity and is mostly employed to assemble car doors. Hemming is a fairly delicate process since it does not only suffice a technical requirement โ€“ to join two panels together โ€“ but also an esthetical one, since the joint panels are an integral part of the vehicle design and as such an important selling point. An unpredicted rupture or an advanced degradation condition of the system would lead to a significant loss in the quality of the final product or to a sudden stoppage of the production line. The development of a PHM system for hemming devices would hence provide a significant advantage, especially if designed to work for both new and legacy equipment. In this paper, we provide the results of a preliminary analysis of a new PHM framework for robotic roller hemming studied to work without having access to PLC data; the employed data-driven methodology is detailed and applied to the case of increasing wear in the head finger roll. Results from different prognostics routines are hence presented and compared.

Publication Year: 
2020
Publication Volume: 
11
Publication Control Number: 
001
Page Count: 
13
Submission Keywords: 
Industrial robot; data driven; LSTM; particle filtering
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Modeling and simulation
Systems and platform applications
Submitted by: 
  
 
 
 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed