<p>Digital twin (DT) technology plays a significant role in advancing smart manufacturing by enhancing reliability and control while reducing human interaction in production processes. It maintains a continuously up-to-date digital representation of physical world entities and enables precise optimization, prediction, and decision-making throughout the fabrication process. However, current limitations in DT integration and limited practical applications remain evident weaknesses. To address the identified gaps, a DT-driven decision support approach consisting of optional decision modes and decision-making algorithms is proposed. A DT architecture is introduced on the basis of global and local DT. The DT intermediary, composed of service-based local DT control devices, is developed to integrate the DTs and implement the optional decision mode into practice. Consequently, near real-time disturbance response and decision execution in the production process are realized while improving computational efficiency. Algorithms of local collaboration and global scheduling are devised to guarantee the stable operation of the production process. To validate the approach’s performance, a case study is conducted in a mixed-model assembly line.</p>