This study addresses multi-objective conflicts in collaborative supply chain inventory optimization by proposing a hybrid framework that integrates XGBoost with NSGA-III. The framework utilizes XGBoost to develop a high-precision demand forecasting model, which is then combined with NSGA-III to achieve tri-objective optimization, encompassing cost, service level, and inventory turnover rate. Implementation in an electronics manufacturing enterprise resulted in an 18.3% reduction in inventory holding costs, an increase in the order fulfillment rate to 96.7%, and a decrease in inventory turnover days to 22. The algorithm utilizes XGBoost's feature importance analysis to guide NSGA-III's search process, resulting in a uniformly distributed Pareto front solution set after 150 iterations. Experimental results demonstrate that the framework enhances the convergence metric of the solution set by 32% compared to NSGA-II in addressing many-objective optimization problems, while achieving a 40% improvement in computational efficiency. This research proposes an efficient integrated approach for supply chain multi-objective collaborative optimization by combining data-driven methodologies with intelligent algorithms.
Cuiting Li (Fri,) studied this question.