Material delivery optimization for make-to-order reconfigurable job shops using an improved chaotic multi-verse algorithm

Abstract

The increasing demand for product customization has highlighted the importance of make-to-order (MTO) material delivery. Although manufacturers have deployed intelligent reconfigurable job shops equipped with flexible workstations and automated guided vehicles (AGVs), challenges remain due to inefficient material scheduling, delayed deliveries, and the complexity arising from diverse material types. This study proposes an active delivery strategy based on a workshop material supermarket, in which both AGV path planning and workstation layout are jointly optimized in response to dynamically changing orders. A multi-objective delivery path model is formulated to support demand splitting while minimizing material delivery costs and maximizing timeliness satisfaction. The model incorporates constraints related to AGV capacity, path feasibility, and demand alignment. To address the nonlinearity and complexity of the problem, an improved chaotic multi-verse optimizer (ICMVO) is proposed. The algorithm employs chaotic encoding to enhance population diversity and mitigate premature convergence. It further integrates gravitational and collision operators to improve global and local search capabilities and adopts adaptive orbital dynamics control to balance exploration and exploitation. A dual-population iterative strategy is employed to enable joint decision-making on workstation coordinates, path direction, and vehicle assignment. Through comprehensive comparisons with state-of-the-art meta-heuristics, the superiority of the ICMVO algorithm and the effectiveness of its components are demonstrated. Moreover, the proposed material delivery optimization method is implemented in a cloud–edge–terminal system and validated in practical MTO reconfigurable job shops through improvements in productivity and cost efficiency.