Multimodal machine learning framework for material selection in injection molding industry

The design phase in injection molding is critical, as it directly influences product quality, production efficiency, and overall manufacturing costs. This paper introduces a multimodal machine learning framework to predict product quality during the design phase, whether for new products or modifications of existing ones. The framework integrates regression, classification, and probabilistic modeling to evaluate material characteristics and process parameters, providing dimensional and mechanical performance predictions for defined products. This methodology minimizes defect risks, optimizes process parameters, and refines production systems. The results highlight the framework’s potential to enhance efficiency, reduce waste, lower costs, and facilitate smarter, data-driven manufacturing aligned with Industry 4.0 principles.