基于条件场景的不确定性石化供应链优化方法

针对产品需求及其价格存在不确定性的石化供应链计划层最优化问题，本文建立了一种基于条件场景的石化供应链最优化方法。用多个离散场景近似随机变量概率的连续分布，根据随机变量的概率分布特征，对场景发生的概率进行参数估计，进而建立了基于场景的两阶段混合整数线性规划(MILP)模型。利用基于场景的优化结果随离散网格数增加而逐渐趋近连续的随机优化结果这一规律，给出了获得最佳离散网格数的方法，实现了计算时间成本与计算精度之间的平衡。在此基础上引入条件概率方法，利用两个随机变量间的相关性，建立了以产品价格及其需求量为不确定性的石化供应链优化方法。结果表明，与传统未考虑随机变量间相关性的一般场景划分方法相比，本文基于条件场景的随机优化方法可以更快地获得最佳场景数目，进而有效降低了计算量。

Conditional scenario based approach to the optimization of petrochemical supply chain with uncertainty

A conditional scenario based approach was proposed for the optimization of tactical level petrochemical supply chain with uncertain product demands and their prices. Based on the probability distributions of the stochastic variables, estimation on the parameters for the probability of each scenario was conducted, and a series of discretized scenarios were introduced to approximate the continuous optimization problem, and a scenario based two-stage mixed integer linear programming (MILP) model was formulated. Result of this optimization tended to approach the continuous optimization when the number of scenario increased, and based on this rule, a method to find the optimal number of scenarios was proposed, so balance between the calculation accuracy and computation time was achieved. Conditional scenario method was further introduced with regard to the correlation of two stochastic variables. Results showed that, compared with traditional method with no variables correlated, the conditional scenario based stochastic optimization method could find the optimal number of scenarios more quickly, and effectively reduce the computational effort.