Skip to Main Content
When executing a manufacturing process, under some circumstances nonconformances may occur in the part, which result in problems such as worker stress, line stoppage, delay in product shipping, and customer loss. In order to reduce or eliminate the above problems, when a nonconformance is identified, it is very important that the possible causes are determined and their solutions put into practice and as soon as possible. In this context, an aspect that should be considered is that in most manufacturing processes the problems to be solved tend to be recurrent, repeating with small variations in relation to its original version, and in this way previous solutions can also be reapplied with small modifications. In this context, this paper presents the implementation of an architecture that combines the Case-Based Reasoning and Multi-Agent approaches in a distributed environment for rapid nonconformance problem solving. The process considered in this work is thermoplastic injection molding, in which the nonconformances occur due to the complex interrelation among the constructive elements of the molded part, the mold, the rheological characteristics of the resin, and the processing conditions, which frequently cause the task of discovering the source of the nonconformance and its respective solution to become slow and costly. A special emphasis in this paper is given to describing the procedure for calculating the similarity measures in the CBR methodology. The developed prototype was based on the Java Agent DEvelopment framework (JADE), having the JColibri CBR framework as core engine.