FUZZY MODELS TO PERFORM CONTAINER INSPECTION AND CLOSING MACHINE PREDICTIVE CONTROL

摘要

The authors have been involved in developing an automated inspection system, based on machine vision, to assess the seaming quality in metal containers (cans) for fishing food. This paper presents a fuzzy model building to make the pass/fail decision for each can, and predict the closing machine adjustment state after closing each can, from the information obtained by the vision system. The can seaming process, and the present, conventional method for quality control of can seaming, are firstly described. Then, the fuzzy modeling, that includes the generation of representative input-output data sets, the used neuro-fuzzy modeling techniques, and the built model is shown. Finally, the performance of the models is tested.