Title: Knowledge driven sensor placement in multi-station manufacturing processes

Authors: Prerna Tiwari; Manoj K. Tiwari

Addresses: Department of Metallurgical and Materials Engineering, Indian Institute of Technology Madras, Chennai, 600036, India ' Department of Industrial Engineering and Management, Indian Institute of Technology, Kharagpur, West Bengal, 721302, India

Abstract: This paper presents a novel methodology for sensor placement in case of multi station manufacturing processes to reduce the dimensional variation in the manufactured product. The proposed methodology integrates knowledge about sensor placement problem with data mining methods. In this paper, sensor placement is termed as design alternative and a set of possible sensor locations as design space. The proposed methodology has four basic steps, i.e.: 1) uniform selection of design alternatives from design space (a set of all possible sensor locations/measurement points); 2) selection of computationally simpler feature functions which can characterise the goodness of a sensor placement; 3) construction of design library based on clustering approach; 4) classification of alternatives present in design library to form design selection rules. An assembly example (i.e., side frame of a sports utility vehicle) has been considered to illustrate the methodology. Moreover, efficiency of the proposed knowledge driven methodology is demonstrated by comparing it with exchange algorithm and other random search approaches which are predominantly used for sensor placement problem.

Keywords: multi-station manufacturing processes; clustering; classification; sensor placement; data mining; sensor locations; design space; design alternatives; automotive assembly; side frame assembly; sports utility vehicles; SUVs.

DOI: 10.1504/IJIEI.2014.066202

International Journal of Intelligent Engineering Informatics, 2014 Vol.2 No.2/3, pp.118 - 138

Received: 04 Jun 2013
Accepted: 19 Jun 2013

Published online: 08 Dec 2014 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article