A Data-Driven Distributionally Robust Optimization Approach for the Core Acquisition Problem

Problem definition: Due to complex electrochemical reactions and physical conditions, the quality of used products (cores) is highly uncertain. The remanufacturer needs to make the acquisition decision under the quality distributional ambiguity. The perfect quality distribution of cores cannot be known to the remanufacturer in practice. We focus on mining the value of multisource data features to identify robust acquisition decisions.Methodology/results: We apply the robust optimization approach integrated with an information fusion technique that can accommodate empirical and public quality information. We develop distributionally robust optimization models based on phi-divergence measures and the imprecise Dirichlet model (DRO-IDM) to derive robust decisions. First, we find that the bounds of the quality probability intervals are identified solely based on the collected data by introducing the imprecise Dirichlet model. We show that the derived finite sample boundary can effectively reduce the scope of the uncertainty set and avoid the no-direction search issue. Second, our models can (i) hedge against distributional uncertainty, (ii) reduce the probability of a robust solution that deviates from the optimal solution, and (iii) correct bias in the acquisition decisions. Third, the DRO-IDM can be extended to create two data-driven models, which can reassess the value of multisource quality information to improve the estimation accuracy of the core quality and maximize the remanufacturer's profit.Managerial implications: Our study provides new insights for remanufacturers: (i) the new remanufacturing process proposed in our work can assist remanufacturers in fully utilizing the values of cores without disassembly; (ii) the efficient information-aware algorithm proposed in our work can be integrated into a smart remanufacturing system that can accelerate digitalization for the modernized remanufacturing sector; (iii) our proposed approach can provide a manner of integrating information fusion and distribution learning into remanufacturing under quality ambiguity