Sensor Optimization for Fault Diagnosis in Single Fixture Systems: A Methodology

Fixture fault diagnosis is a critical component of currently evolving techniques aimed at manufacturing variation reduction. The impact of sensor location on the effectiveness of fault-type discrimination in ?;uch diagnostic procedures is signijcant. This paper proposes a methodology for achieving optimal fault-type discrimination through an optimized conjguration of dejned "sensor locales. " The optimization is presented in the context of autobody fucturing-a predominant cause of process variability in automobile assembly. The evaluation criterion for optimization is an improvement in the ability to provide consistency of best match, in a pattern recognition sense, of any fucture error to a classijied, anticipated error set. The proposed analytical methodology is novel in addressing optimization by incorporating fucture design specijications in sensor locale planning-constituting a Design for Fault Detectability approach. Examples of the locale planning for a single fucture sensor layout and an application to an industrial fucture conjguration are presented to illustrate the proposed methodology.