Providing a Framework for a Sustainable Supply Chain in Plant-Based Milk Production Based on the Internet of Things

This research designs a framework for a sustainable supply chain in plant-based milk production utilizing the Internet of Things (IoT). The study aims to develop a multi-objective model for the supply chain, leveraging IoT tools to monitor and optimize internal network processes. Within this system, IoT technology is employed to track inventory levels at distribution centers, preventing product spoilage and enhancing the distribution network. When inventory levels reach a predefined threshold, the ordering process is intelligently executed using IoT. Additionally, this process incorporates intelligent routing within the distribution and logistics network, driven by data collected from IoT devices, as well as production planning and the supply of raw materials and packaging. To address the complex challenges of this supply chain, the GAMS software and a hybrid NSGA-II algorithm are utilized. The findings of this research demonstrate that the proposed hybrid algorithm (combining the NSGA-II algorithm with initial solutions derived from the MOSA algorithm) yields superior solutions, albeit with a longer computation time compared to the standalone NSGA-II algorithm. While the hybrid algorithm provides better solutions, its problem-solving duration is extended due to additional processes. Specifically, the MOSA algorithm is first run independently to generate initial solutions, which requires extra time. These initial solutions are then fed into the NSGA-II algorithm, where evolutionary operations refine them to produce a set of optimal solutions. This added step introduces greater complexity and computational time. In contrast, the standalone NSGA-II algorithm, without the need for generating initial solutions or additional processes, offers a shorter solution time. However, the hybrid algorithm, by leveraging the initial solutions from the MOSA algorithm, achieves more optimal results. These initial solutions serve as a better starting point for exploring the solution space, enabling faster convergence to optimal regions. This improvement in solution quality highlights the effectiveness of combining these two algorithms, particularly for complex, multi-objective problems.