This paper presents a deep-reinforcement-learning-based digital twin framework for manufacturing process optimization, focusing on plastic injection molding as a practical use case. The proposed system combines supervised learning and deep reinforcement learning (DRL) to enable autonomous process control. The digital twin continuously updates the virtual representation of the system using real-time data and makes intelligent decisions to optimize operational metrics. The framework improves product quality while reducing costs and minimizes human involvement. The study demonstrates how the digital twin technology enhances productivity and effectiveness in production processes, advancing its application in the industrial sector. The digital twin is designed as a specific-purpose system, tailored to achieve particular objectives such as improving product quality or optimizing manufacturing parameters. This approach enhances the feasibility and value of digital twins in diverse contexts. The system includes data ingestion and processing, supervised learning model training, real-time prediction of production quality, decision-making using DRL, and optimization of processing settings. The digital twin also integrates traceability systems for quality assurance and feedback loops for continuous improvement. The paper reviews existing literature on digital twin technologies and their applications in various fields, including healthcare, maintenance, manufacturing, and smart cities. It highlights the growing interest in digital twins for process optimization, particularly in the automotive industry. The study also addresses the challenges of implementing digital twins, such as the complexity of modeling manufacturing operations and the need for adaptive solutions. The research contributes to the field by introducing a data-driven approach that captures the complex behavior of manufacturing systems through real-time data. It fills gaps in decision-making aspects of digital twins by integrating AI and machine learning for continuous improvement. The proposed framework is validated through a real-world case study in plastic injection molding, demonstrating its effectiveness in optimizing process parameters and achieving high-quality outputs. The study shows how digital twin technology can improve the efficiency and adaptability of manufacturing processes, offering a novel and practical solution for industrial applications.This paper presents a deep-reinforcement-learning-based digital twin framework for manufacturing process optimization, focusing on plastic injection molding as a practical use case. The proposed system combines supervised learning and deep reinforcement learning (DRL) to enable autonomous process control. The digital twin continuously updates the virtual representation of the system using real-time data and makes intelligent decisions to optimize operational metrics. The framework improves product quality while reducing costs and minimizes human involvement. The study demonstrates how the digital twin technology enhances productivity and effectiveness in production processes, advancing its application in the industrial sector. The digital twin is designed as a specific-purpose system, tailored to achieve particular objectives such as improving product quality or optimizing manufacturing parameters. This approach enhances the feasibility and value of digital twins in diverse contexts. The system includes data ingestion and processing, supervised learning model training, real-time prediction of production quality, decision-making using DRL, and optimization of processing settings. The digital twin also integrates traceability systems for quality assurance and feedback loops for continuous improvement. The paper reviews existing literature on digital twin technologies and their applications in various fields, including healthcare, maintenance, manufacturing, and smart cities. It highlights the growing interest in digital twins for process optimization, particularly in the automotive industry. The study also addresses the challenges of implementing digital twins, such as the complexity of modeling manufacturing operations and the need for adaptive solutions. The research contributes to the field by introducing a data-driven approach that captures the complex behavior of manufacturing systems through real-time data. It fills gaps in decision-making aspects of digital twins by integrating AI and machine learning for continuous improvement. The proposed framework is validated through a real-world case study in plastic injection molding, demonstrating its effectiveness in optimizing process parameters and achieving high-quality outputs. The study shows how digital twin technology can improve the efficiency and adaptability of manufacturing processes, offering a novel and practical solution for industrial applications.