Multiobjective Optimization-Aided Decision-Making System for Large-Scale Manufacturing Planning

Area of Science:

• Operations Research
• Industrial Engineering
• Computational Optimization

Background:

• Real-world manufacturing planning (MP) involves complex challenges, including numerous decision variables and conflicting objectives.
• Existing methods struggle with the large scale and difficult constraints typical of practical MP tasks.
• Balancing order fulfillment rates and total costs requires sophisticated optimization techniques.

Purpose of the Study:

• To develop an interactive multiobjective optimization-based system for real-world manufacturing planning.
• To address the challenges of maximizing order fulfillment and minimizing costs simultaneously.
• To provide decision-makers with a tool for achieving satisfactory trade-offs without excessive computation.

Main Methods:

• The manufacturing planning task is modeled as a multiobjective integer programming (MOIP) problem.
• A novel two-stage multiobjective optimization algorithm (TSMOA) is introduced to solve the MOIP.
• TSMOA decomposes the MOIP into a series of single-objective problems (SOPs) and employs a new SOP solving strategy for efficiency.

Main Results:

• The developed MP system, incorporating TSMOA, demonstrates efficiency in real-world manufacturing planning applications.
• TSMOA effectively reduces computational cost by approximating complex SOPs with sequential, simpler ones.
• Validation on benchmark problems confirms the effectiveness and promising performance of TSMOA.

Conclusions:

• The interactive multiobjective optimization system offers a practical solution for complex manufacturing planning.
• TSMOA provides an efficient and effective approach to handling large-scale, multiobjective optimization problems in manufacturing.
• The proposed system and algorithm show significant potential for improving manufacturing planning processes.