Deep Reinforcement Learning for Solving the Flexible Flow Shop Scheduling Problem

Wen Xinran

Abstract

his paper presents a deep reinforcement learning algorithm applied to address the Flexible Flow Shop Scheduling Problem (FFSP) with the objective of minimizing the makespan. The FFSP is a challenging combinatorial optimization problem commonly encountered in manufacturing and production environments. Traditional optimization techniques often struggle to handle its complexity and dynamic nature. In this work, Proximal Policy Optimization (PPO), a state-of-the-art deep reinforcement learning technique, is explored to efficiently and effectively solve the FFSP. Through experimentation and evaluation on a real-life instance, this algorithm shows superiority in achieving better performance compared to traditional optimization methods including exact algorithm and certain heuristic algorithms.