A MODELING ERROR APPROACH TO PARTS QUALITY ASSESSMENT

摘要

This investigation focuses on the formulation and assessment of a strategy for the discriminationrnof good and bad manufactured parts. The proposed approach is based on an exogenous autoregressivernstructural model of the good parts and on the error of this model in predicting the measured responses ofrngood and bad parts. The discrimination strategy relies on the expectation that the modeling error of goodrnparts will be smaller than the one of bad parts since the model is representative of one or several goodrnparts. Six techniques were then introduced to reduce the modeling error, a time-dependent vector, into arnfew overall discrepancy measures which quantify in different ways the ?magnitude? of the error vectorrnand thus can each be used as a basis for the discrimination of the good parts from the bad ones. Thernresults of a preliminary assessment with simple one and four degree of freedom models of the partsrnclearly indicates that the present error-based discrimination strategy performs very well - it is generallyrneither optimum or close to it.