Supply chain simulation models are widely used for assessing supply chain performance and analyzing supply chain decisions. In combination with derivative-free optimization algorithms, simulation models have shown great potential in effective decision-making. Most of the derivative-free optimization algorithms, however, assume continuity of the response, which may not be true in some practical applications. In this work, a supply chain inventory optimization problem is addressed that results in a discontinuous objective function. A derivative-free optimization framework is proposed that addresses the discontinuities in the objective function. The framework employs a sparse grid sampling and support vector machines for identification of discontinuities. Computational comparisons presented show that addressing discontinuity leads to more cost-effective decisions over existing approaches.

中文翻译：

---

多企业供应链的不连续无导数优化框架

供应链仿真模型被广泛用于评估供应链绩效和分析供应链决策。结合无导数优化算法，仿真模型已显示出有效决策的巨大潜力。但是，大多数无导数优化算法都假定响应是连续的，在某些实际应用中可能并非如此。在这项工作中，解决了导致不连续目标函数的供应链库存优化问题。提出了无导数优化框架，该框架解决了目标函数中的不连续性。该框架采用稀疏网格采样和支持向量机来识别不连续性。