Optimal Decision-Making in Production Processes Based on Sequential Sampling and Dynamic Programming

Abstract

Aiming at the quality control and cost optimization problems in the production process of electronic products, this paper proposes a complete decision-making framework based on sequential sampling and multi-stage dynamic programming. First, the sequential probability ratio test (SPRT) is used to design a dynamic sampling inspection scheme with minimal detection times, which can effectively identify unqualified parts and reduce the cost of quality inspection. Second, a multi-stage dynamic programming model is established to optimize the decisions of procurement, inspection, assembly and after-sales treatment. Third, the model is extended to the complex production network with multiple processes and multiple parts based on Markov decision process (MDP). The proposed framework can adapt to the uncertainty of the production environment and the variability of product types, which improves the stability and reliability of the whole production system. Meanwhile, this method can realize the real-time transmission and intelligent analysis of production data, and support the intelligent transformation of traditional manufacturing. The experimental results show that the proposed method can reduce the total production cost by 18%–25% and control the product defect rate below 5%. In addition, this model can effectively shorten the production cycle and improve the utilization rate of production resources. The framework has high efficiency and strong practicability, which can provide scientific support for production decision-making of manufacturing enterprises.