Faults and failures prediction in injection molding process

In production of the polymeric parts, injection molding is an important processing technique which provides easy automation and economic manufacturing. Since several parameters indicate crucial influences on this method, artificial intelligence (AI) approaches have been utilized to optimize the injection molding process. In this study, an intelligent system is implemented to detect different faults in injection molding. To this aim, we used the fuzzy case-based reasoning (fuzzy CBR) approach as a complementary reasoning method in AI. CBR solves new problems via referring to the nearest solutions of the most similar cases. Problems in which attribute values have fuzzy characteristics are fuzzified and similarity measurements developed with respect to these features. Using fuzzy logic in the retrieval phase of our CBR system leads to easier transfer of knowledge across domains. In the current research, the triangular fuzzy numbers are utilized to represent the imprecise numerical quantities in the relationship values of each feature and related parameters based on domain experts' knowledge. An implemented system is evaluated by detection of various faults in a production line. The obtained results proved capability and accuracy of the proposed system in detection of faults. The system is much faster than traditional method and indicates a stable product quality. The proposed system can also be adapted for other complex products in the injection molding process.