Integrated Planning and Machine-Level Scheduling for High-Mix Discrete Manufacturing: A Profit-Driven Heuristic Framework
By: Runhao Liu , Ziming Chen , You Li and more
Potential Business Impact:
Makes factories finish jobs on time, every time.
Modern manufacturing enterprises struggle to create efficient and reliable production schedules under multi-variety, small-batch, and rush-order conditions. High-mix discrete manufacturing systems require jointly optimizing mid-term production planning and machine-level scheduling under heterogeneous resources and stringent delivery commitments. We address this problem with a profit-driven integrated framework that couples a mixed-integer planning model with a machine-level scheduling heuristic. The planning layer allocates production, accessory co-production, and outsourcing under aggregate economic and capacity constraints, while the scheduling layer refines these allocations using a structure-aware procedure that enforces execution feasibility and stabilizes daily machine behavior. This hierarchical design preserves the tractability of aggregated optimization while capturing detailed operational restrictions. Evaluations are conducted on a real industrial scenario. A flexible machine-level execution scheme yields 73.3% on-time completion and significant outsourcing demand, revealing bottleneck congestion. In contrast, a stability-enforcing execution policy achieves 100% on-time completion, eliminates all outsourcing, and maintains balanced machine utilization with only 1.9 to 4.6% capacity loss from changeovers. These results show that aligning planning decisions with stability-oriented execution rules enables practical and interpretable profit-maximizing decisions in complex manufacturing environments.
Similar Papers
Automatic Operator-level Parallelism Planning for Distributed Deep Learning -- A Mixed-Integer Programming Approach
Machine Learning (CS)
Makes big AI models train faster and smarter.
Automated Constraint Specification for Job Scheduling by Regulating Generative Model with Domain-Specific Representation
Artificial Intelligence
Helps factories plan production automatically and faster.
Hierarchical Planning and Scheduling for Reconfigurable Multi-Robot Disassembly Systems under Structural Constraints
Robotics
Robots take apart things without breaking them.