Dynamic Inventory Optimization and Cost Reduction Using Motif Knowledge Heterogeneous Dollmaker Graph Attention Network for Global Supply Chain Efficiency

Logical inventory control is vital in maintaining production flow and reduction in the costs of operation in the manufacturing chains worldwide. The classic fixed models tend to strike a balance between the safety stock levels, reorder points and order quantities in the dynamic conditions of demand resulting in either the overstock or the necessity to spend a lot of money on the line stoppages. In order to overcome them, this work suggests a dynamic inventory management strategy that relies on the Motif Knowledge Heterogeneous Dollmaker Graph Attention Network (Mot-KHD-2GAN), which is aimed at optimizing safety stock, reorder levels, and order quantities of every inventory component concurrently. The strategy uses the data of Distribution hubs in North America, Europe and Asia-Pacific where Economic order quantity (EDOQ) model had cut down costs significantly USD 23,000 in North America, USD 16,500 in Europe and USD 21000 in Asia-Pacific averaging 17 percent cut down in inventory spending all over the globe. Moreover, the inventory turnover level has increased dramatically in all regions, which is indicative of the increased supply chain flexibility and responsiveness. The pre-processing is done by Reversible Automatic Selection Normalization (RASN) to guarantee the data consistency and eliminate the distributional bias and the features are extracted by an Empowering Decision Transformer (EDT) to extract the correlations between the contextual cost dynamics and inventory patterns. The Mot-KHD-2GAN is composed of a knowledge graph and graph neural network (KG-GNN), a Motif-Based Heterogeneous Graph Attention Network (MBHAN), and the parameters are optimally adjusted with the support of a Dollmaker Optimization Algorithm (DOA), ensuring that there is optimal convergence. The proposed approach has been tested for a prediction accuracy of 99.9%, which confirms the efficiency of the approach to minimize the inventory cost and avoid interruption of production. The framework has major benefits, such as adaptive demand variability learning across regions and better optimization of the inventory policies to improve the global supply chain performance.