<![CDATA[This research investigates the integration of multi-objective optimization and reinforcement learning (RL) to enhance decision-making within a Decision Support System (DSS), with a focus on dynamic manufacturing environments. The primary objective is to optimize decision quality by balancing three conflicting objectives: minimizing production costs, maximizing efficiency, and minimizing risk, while adapting to real-time changes in demand. The research employs a hybrid approach, combining static optimization to compute Pareto-optimal solutions with RL to enable the system to learn from feedback and improve over time. The research design involves developing a mathematical model that integrates both techniques, followed by a numerical example to test its effectiveness in balancing the objectives. The methodology includes formulating cost, efficiency, and risk functions, solving the multi-objective optimization problem, and implementing a Q-learning-based RL algorithm to refine decision-making based on real-time data. The model was tested using time-dependent demand to simulate a realistic production environment. The main results demonstrate that the hybrid model effectively balances conflicting objectives, with the RL component adapting production decisions to fluctuating market conditions. The system identified an optimal production level around x=60 units, offering a balance between cost, efficiency, and risk. The findings highlight the model's capability to enhance decision-making adaptability in dynamic environments compared to traditional static approaches. In conclusion, this research provides a novel method for improving decision quality in DSS by integrating multi-objective optimization and RL, offering valuable insights for industries requiring adaptive, real-time decision-making. Future research could extend this model to more complex environments and explore its scalability in larger, real-world applications]]>
