Integrated scheduling of distributed assembly hybrid flow shops and equipment operations in dual-distributed networks

Efficient spare parts supply is constrained by the ‘dual-distributed’ manufacturing structure and the lack of bidirectional coordination with equipment operational strategies. To bridge these gaps, this paper investigates a novel integrated scheduling problem (ISP-DPTAO). Extending the classical distributed assembly hybrid flow shop context, our model jointly optimises production, transportation, assembly, and active demand-side operation strategies. We formulate a mixed-integer linear programming model to minimise bilateral costs. To solve this highly coupled problem, an Adaptive Policy-driven Memetic Algorithm (APMA) featuring a two-stage cooperative search architecture is proposed. Unlike static mechanisms, APMA's second stage integrates a reinforcement learning-based network. By extracting environmental state features, this network adaptively selects the most promising neighbourhood operators to refine elite solutions, enabling intelligent local exploitation. Extensive experiments and benchmarks demonstrate the superiority of APMA. Comparative analyses empirically validate the economic necessity and effectiveness of the proposed bilateral collaborative mechanism.