With the growing demand for market personalization, matrix manufacturing has emerged as a new solution due to its integrated and collaborative characteristics, offering a quick response to market changes. In high-flexibility matrix manufacturing workshops, production and logistics are highly coupled and interact dynamically, which introduces significant complexity to the system. However, traditional production models treat production and logistics as independent modules, often leading to synchronization issues. This misalignment causes problems such as resource waiting and inventory buildup. These challenges highlight the limitations of traditional models, which are no longer suitable for matrix-based manufacturing environments. This paper studies the production-logistics collaborative scheduling problem in matrix manufacturing workshops, aiming to minimize makespan and waiting time. A mixed-integer programming model is proposed and solved using a dual population collaborative genetic algorithm (DCGA). The algorithm incorporates dual-layer encoding for production task sequencing and machine selection and enhances optimization through population collaboration. Case studies and experiments demonstrate that the proposed model and DCGA outperform traditional methods in complex scheduling environments.