Improved SAC Algorithm for Solving the Interactive Hybrid Disassembly Line Balancing Problem Considering Multiskilled Workers

Abstract

With the advancement of manufacturing and the progression towards intelligent production, more and more production enterprises are focusing on the hybrid disassembly line balancing problem to enhance efficiency and reduce costs. This work presents an interactive hybrid disassembly line balancing problem considering multiskilled workers and its corresponding mathematical model. Additionally, we incorporated the concept of carbon emissions into the problem to ensure the green sustainability of the production process. We employ an improved soft actor-critic (SAC) algorithm for solving the problem to achieve the goals of maximizing profit, minimizing idle time, and minimizing carbon emissions. The experimental results demonstrate that the improved SAC algorithm performs well on six different experimental scale cases. Specifically, multiobjective soft actor-critic (MO-SAC) produces solutions that are closer to the Pareto front than those obtained by deep deterministic policy gradient and nondominated sorting genetic algorithm II. This advantage is particularly pronounced in medium and large-scale problem instances. Statistical comparisons using nondominated solution metrics (spread, epsilon, and IGD+), Pareto front distributions, and computational time show that MO-SAC achieves better convergence and maintains comparable computational efficiency, highlighting its performance benefits in complex scenarios. The results verify the effectiveness of our proposed approach, highlighting the superiority of the improved SAC algorithm in addressing the multiobjective interactive hybrid disassembly line balancing problem.