A Quantum-Inspired Archimedes Optimization Algorithm for Hybrid-Load Autonomous Guided Vehicle Scheduling Problem
Binghai Zhou,Lingwei Zhao
DOI: https://doi.org/10.1007/s10489-023-04962-1
IF: 5.3
2023-01-01
Applied Intelligence
Abstract:Facing the two major challenges of just-in-time (JIT) and energy-saving in part-feeding system for mixed-model assembly lines in automotive industry, this paper focuses on optimizing the part-feeding process by investigating the autonomous guided vehicles (AGV) routing and scheduling problem. A hybrid feeding policy, called dual-distribution, is proposed, which considers the utilization of different types of AGVs in various part-feeding policies. The problem is formulated as a mixed-integer programming model with the objectives of simultaneously minimizing the line-side inventory and AGV energy consumption in the part-feeding system. To solve this problem, a Q-learning-based multi-objective quantum-inspired Archimedes optimization algorithm (QMQAOA) is developed with a customized encoding and decoding approach to solve the problem. Besides, the quantum rotation gate initialization mechanism, the Q-learning-based neighborhood search strategy, and the neighboring distance calculation are integrated into the algorithm to improve both solution quality and convergence rate. Finally, numerical experiments are conducted to evaluate the performance of QMQAOA by comparing it with the Gurobi solver and other benchmark algorithms. The results demonstrate the superiority of QMQAOA, with performance superiority rates of 90/90, 77/90, and 90/90 achieved for degree of Pareto optimality (DPO), evenness of solutions (ES), and inverted generational distance (IGD) indicators, respectively. From managerial insights, the application of different types of AGVs in the part-feeding system is shown to enhance efficiency when compared to using a single type of AGV in several instances. These findings provide valuable insights for optimizing the automotive part-feeding system.