Author:
Wa-Muzemba Tshibangu
Affiliation:
Morgan State University, United States
Keyword(s):
Robust Design, Optimization, Taguchi Method, Compromise Programming, Fms, Simulation.
Related
Ontology
Subjects/Areas/Topics:
Industrial Engineering
;
Informatics in Control, Automation and Robotics
;
Manufacturing Systems Engineering
;
Performance Evaluation and Optimization
;
Resources and Knowledge Management in Industry
;
Systems Modeling and Simulation
Abstract:
Competitive advantage of a firm is usually reflected through its superiority in production resources and
performance outcomes. In order to achieve high performance (e.g., productivity) and significantly improve
product quality, major US industries have promoted and implemented Robust Design (RD) techniques
during the last decade. RD is a cost-effective procedure for determining the optimal settings of the control
factors that make the product performance insensitive to the influence of noise factors. In this research, we
employ and compare two RD optimum-seeking methods to optimize a flexible manufacturing system
(FMS). Taguchi Method (TM), which uses robust design concept, i.e., Signal-To-Noise Ratio (S/N) to
reduce the output variation, is applied first. Taguchi’s approach to robust design drawn much criticism
because it relies on the signal-to-noise (S/N) ratio for the optimization procedure. Because of this paramount
criticism, a second method known as the Compromise Programming
(CP) approach, i.e., the weighted
Tchebycheff, is also used. This method formulates the robust design as a bi-objective robust design (BORD)
problem by taking into account the two aspects of the RD problem, i.e. minimize the variation and optimize
the mean. This approach seeks to determine the RD solution which is guaranteed to belong to the set of
efficient solutions (Pareto points). Both methods use a RD formulation to determine an optimal and robust
configuration of the system under study. The results gained through simulations and analytical formulations
show that the current ways of handling the multiple aspects of the RD problem by using Taguchi’s S/N ratio
may not be adequate.
(More)