Enhancing parallel assembly sequence planning (PASP) with an advanced priority relationship algorithm (APRA) using machine learning

This study introduces an advanced approach to the Parallel Assembly Sequence Planning (PASP) problem, addressing the complexities of task interdependencies and real-time adaptability across multiple parallel assembly lines. Traditional PASP methods often rely on static heuristics and struggle to accommodate dynamic production environments. To overcome these limitations, we propose the Advanced Priority Relationship Algorithm (APRA), a machine learning-enhanced framework that dynamically predicts assembly sequence parameters using a random forest model. APRA incorporates a priority relationship algorithm to optimize task sequencing in real-time, striking a balance between efficiency, cost, and quality. Experimental evaluations on complex assembly scenarios demonstrate APRA’s effectiveness, achieving a 26% reduction in assembly time (from 120 to 88.8 h) and a Coefficient of determination ( R 2 ) accuracy score of 0.8686, capturing 86.86% of the variability in efficiency. Additionally, small-scale validation confirms its ability to provide near-optimal solutions with reduced computational time, making it feasible for large-scale industrial applications. This research presents a novel problem formulation for PASP and a robust, adaptive solution framework, providing manufacturers with a powerful tool for enhancing productivity and responsiveness in dynamic assembly scheduling.