Gelişmiş plastik enjeksiyon kalıplama ve endüstriyel uygulama için ML odaklı biliş

For a long time, plastic injection molding has been an important part of mass production in many fields. But in today's competitive market, traditional molding methods don't always work well when you need high efficiency and quality that stays the same. We need better and smarter ways to make things because of the need for more sustainable practices, the use of more recycled materials, shorter production times, and rising labor costs. This thesis describes a cognition-based method that is meant to work without depending on the shape of the parts, the type of material, or the specific production equipment used in the injection molding process. To do this, cavity pressure sensors were carefully put in key parts of the mold to collect data that was unique to each cycle. This information helped us find a stable operational range, or "reliable zone," that guarantees that high-quality parts are always made. One important thing this work does is show that changes in the cavity pressure curve are related to both the quality of the parts and the settings of the machine. Using this connection, a convolutional neural network (CNN) was trained to create a basic knowledge system that helps operators by suggesting ways to fix problems when they notice that something is wrong. The model that was made was able to correctly classify 98% of the time. After the baseline was made, the proposed method was used on two real-world case studies that involved parts with different shapes, materials, and quality standards. For each application, a method for adapting knowledge to specific tasks was used, and when the system was put into real-world production environments, it had an average accuracy rate of 95%.