Quantitative Estimation of Fuzzy Classifier Applying to the Quality Control in Plastics Processing

摘要

An effort of quality assurance of plastics processing leads to appearances of a lot of effective methods aimed at On-line quality monitoring of finished products. As a very useful model-free instrument, fuzzy cluster analysis can be applied to the on-line quality monitoring of the plastics processing. But the precisely determination of α-levels, which are used to hierachicaly classify the samples in fuzzy cluster analysis, runs into special difficulty in the applications, mainly because their determination often bases on the "experience". As an optimization tool, golden section search method conduces to quantitative determination of a-levels, so that the performance of the fuzzy classifier can be improved.