Work

A Dual-Stream Deep Reinforcement Learning Framework for Hot Rolling Production Scheduling

Year: 2026

Type: article

Abstract: Hot Rolling Production Scheduling (HRPS) is a crucial combinatorial optimization problem characterized by severe conflicts between rigid physical rolling rules and strict order due dates. While real-time scheduling is essential for dynamic manufacturing, traditional meta-heuristics suffer from severe computational time bottlenecks. Conversely, standard end-to-end Deep Reinforcement Learning (DRL) models offer rapid inference but typically struggle with spatio-temporal feature entanglement, training instability under extreme penalty landscapes, and poor zero-shot scale generalization. To bridge these gaps, this paper proposes a novel framework named Dual-Stream Group-Optimize Policy Optimization with Multiple Optima (DSGO-POMO). The framework introduces three core innovations: (1) a Dual-Stream intervention network that explicitly decouples and synergistically fuses physical attributes with temporal slacks; (2) a Group Relative Policy Optimization (GRPO) training mechanism to stabilize policy updates; and (3) an Entropy-Aware and Dual-Annealed Differential Active Search (EA-DAS) strategy to seamlessly adapt pre-trained weights to out-of-distribution scales. Extensive computational experiments validate the superiority of the proposed framework. On medium-scale instances (n=50), the basic DSGO-POMO slashes inference time to merely 0.09 min compared to the approximately 46 min required by classical heuristics while simultaneously achieving lower total costs. Furthermore, during zero-shot extrapolation on ultra-large instances (n=200), the framework coupled with EA-DAS outperforms the best heuristic baseline by 14.7% in solution quality and completes the search in under 2 min. These numerical breakthroughs confirm that DSGO-POMO provides a blazing-fast, highly robust, and industrially viable rescheduling paradigm.

Source: Machines

DOI: 10.3390/machines14070741

Authors: Chi Wang, Wang Cao, Min Huang

Institution: Northeastern University

Language: English

Topic: Scheduling and Optimization Algorithms

Subfield: Industrial and Manufacturing Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,896