求解非线性项目组合选择问题的多种线性化技术比较研究

Comparing Multiple Linearization Techniques of Solving Nonlinear Project Portfolio Selection Problem with Interdependency

本文通过现有的线性化技术,将项目组合选择中存在的非线性规划问题转换成等价的整线性规划模型,并利用数值实验比较这些转换后的模型求解效率。为此,本文将现有线性化技术应用于考虑了项目间相互作用关系的多准则项目组合选择问题上,给出了三个等价模型。在此基础上通过五个类别的数值实验分析,较为客观的比较了每个模型的效率。最后根据程序结果得出现有线性化技术的优缺点,以便更好的指导多准则项目组合选择问题的求解。

The purpose of this paper is to reformulate the nonlinear programming model into several equivalent linear integer programming ones by utilizing existed linearization techniques, and efficiency of each method is compared by a well-designed numerical experiment. Existed linearization techniques have been investigated and applied into a multi-criteria project portfolio selection problem(MCPPSP) considering interdependency, where three equivalent models are introduced. Thus, by well-designed numerical experiment with 5 categories of instances,the efficiency of each method can be objectively known. Finally,conclusion about the advantages and deficiencies these linearization techniques can be drawn,and some advisable hints about how to handle MCPPSP are given out.