Production Process Optimization and Defective Rate Control Based on Adaptive Sequential Detection and Monte Carlo Tree Search

Abstract

With the increasing globalization and complexity of supply chains, modern manufacturing faces challenges in ensuring product quality while minimizing inspection costs. Companies must balance the need for reliable components from diverse suppliers with the risks of defective parts that could disrupt production. Traditional sampling methods often lack flexibility, leading to inefficiencies in defect detection. To address these challenges, this study integrates adaptive sequential detection with Monte Carlo Tree Search (MCTS) to optimize production processes. The adaptive detection approach adjusts sample sizes dynamically, ensuring defective parts are rejected with 95% confidence or accepted with 90% confidence, maintaining precision while controlling costs. When the defective rate nears 0.11, the method requires up to 155 samples for accurate decisions. MCTS further enhances decision-making by simulating multiple production strategies and identifying the optimal path, balancing inspection costs with quality control. Among the six tested scenarios, the best-case strategy achieved an expected return of 47.2 units, highlighting the model’s effectiveness. This framework provides a practical solution for manufacturers to improve decision-making under uncertain conditions and can be extended to multi-stage production lines, offering valuable insights for addressing supply chain disruptions and complex operational challenges.