This study first defined two major problems in the automatic inspection of the inspection machine for injection medicines, which were the inspection results falling beyond the target range resulting in a low performance ratio and a lengthy process of machine parameter adjustments to reach the expected performance ratio. It then performed a manual inspection to determine the existing performance status. During the analysis phase of the study, TRIZ and CBR (case based reasoning) were integrated and used as the manifest tool to inspect and adjust the parameters related to the Knapp test for the purpose of utilizing their ways of systematic thinking to find the relative parameters that might affect the inspection quality, and further to confirm four major parameters through screening by TRIZ and CBR, including the magnification, the bi-qualitative alloplasm of the relative positions of different images, the maximum floating foreign particle lumps and the maximum foreign particle lumps. During the improvement phase of the study, a set of test procedures combining the Taguchi method and analysis of variance was proposed to determine the optimum parameter combination of the major parameters. The validation tests confirmed that the problem solution structure developed in this study could allow the automatic inspection to realize the expected efficiency. This structure could make a significant contribution to the application of the Knapp test in medical automatic inspection systems, and serve as a reference to industries relying on manual production or inspection.
