供应链质量与多阶段决策的贝叶斯–混合优化

Bayesian-Hybrid Optimization for Supply Chain Quality and Multi-Stage Decision Making

摘要: 本研究针对制造业供应链质量管理与多阶段生产决策优化问题，提出了一种融合贝叶斯后验更新、Z检验与混合优化算法的创新框架。通过贝叶斯方法动态估计零配件次品率，结合Z检验优化抽样样本量，在保证决策精度的同时减少检测成本。针对多阶段生产流程，构建动态规划与遗传算法结合的混合优化模型，定义半成品/成品次品率传播公式与拆解回收价值模型，实现检测、装配、拆解决策的全局优化。案例研究表明，该方法在8零配件2工序场景中实现最低成本6920.57元，最优方案倾向于选择性检测与拆解回收。创新性体现在次品率动态传播机制、贝叶斯–遗传混合优化框架及全流程成本集成建模，为复杂制造系统提供效率与可持续性并重的决策工具。

Abstract: In this study, an innovative framework integrating Bayesian posteriori updating, Z-test and hybrid optimization algorithms is proposed for the manufacturing supply chain quality management and multi-stage production decision optimization problem. By dynamically estimating the defective rate of spare parts through the Bayesian method, the sampling sample size is optimized by combining with the Z-test, which ensures the decision-making accuracy and reduces the inspection cost at the same time. For the multi-stage production process, a hybrid optimization model combining dynamic programming and genetic algorithm is constructed to define the semi-finished/finished product defective rate propagation formula and the disassembly recycling value model, so as to realize the global optimization of the decision-making of inspection, assembly and disassembly. The case study shows that the method realizes the lowest cost of 6920.57 yuan in the 8-parts-2-processes scenario, and the optimal solution tends to selective inspection and disassembly and recovery. The innovativeness is reflected in the dynamic propagation mechanism of defective rate, the hybrid Bayesian-genetic optimization framework, and the integrated modeling of whole-process cost, which provides a decision-making tool for complex manufacturing systems with both efficiency and sustainability.