Superimposed Poisson Distribution Variable Neighborhood Search for Scheduling of Parallel Multi-track Shuttle Loop System

Abstract

This article presents a layout scenario of parallel multiple tracks shuttle loop system and its scheduling method within an automated storage and retrieval system. Leveraging the motion mathematical model articulated through angular coordinates and track coding, we established a multi-agent simulation environment. The simulation encompasses RGV agents, task assignment agents, track allocation agents, and charging decision-making agents. The actions of individual agents are directly or indirectly guided by a three-tier decision coding method. We also propose a heuristic algorithm, termed Superimposed Poisson Distribution-Variable Neighborhood Search, to optimize system efficiency through Simulation-Based Optimization. Numerical experiments validate the performance of the algorithm using example scenarios of varying scales. Furthermore, we confirm the superior efficiency of parallel multiple track layouts, assess the impact of the number of RGVs on system throughput, and identify the optimal number of RGVs for different track layout configurations. A sensitivity analysis is conducted to explore the effect of physical parameters on system efficiency.