An LLM-assisted optimization algorithm for energy-efficient flexible job shop scheduling problem with unreliable machines and AGVs

Abstract

With technological advancements, the influence of small-batch flexible production models in manufacturing continues to grow, making the flexible job shop scheduling problem a key approach to addressing these industry challenges. Simultaneously, the impact of random machine failures and limited buffers cannot be underestimated, drawing academic attention to scenarios beyond fault-free production lines. Moreover, in real-world multi-process manufacturing, logistics processes are critical. Considering the simultaneous optimization of completion time and energy consumption, this paper proposes a novel energy-efficient flexible job shop scheduling problem with automatic guided vehicles (EFJSP-AGV) considering unreliable machines. To address the EFJSP-AGV, a new mixed-integer linear programming model is developed for the exact solution of small-scale instances. Furthermore, an improved memetic algorithm is proposed, which is enhanced by a heuristic designed with the assistance of large language models (LLMs). The feasibility and effectiveness of the MILP model and the improved algorithm is verified through experimental simulation.