基于混合支持向量机的国防R&D项目中止决策方法

由于RD活动的高风险性,对国防RD项目开展中止决策研究具有重要的理论和现实意义.将传统RD项目中止决策的解决途径分为两类:分类型和回归型.在分析这两种解决途径存在不足的情况下,考虑到国防RD项目自身的特点,提出一种基于混合支持向量机(support vector machine,SVM)的项目中止决策方法;同时,针对模型的建立问题,提出采用交叉验证的方法,通过粒子群优化(particle swarm optimization,PSO)算法实现模型的优化选择.通过建立混合决策模型,可以得出项目中止决策的明确结论.应用分析表明,该方法能够同时实现对项目的分类和排序,较常用方法利用信息更为全面,得出结论更为细致,对实践中的项目中止决策具有较好的适应性.

National Defense R&D Project Termination Decision-Making Method Based on Composite Support Vector Machine

As the high-risk in the R&D activities,it is very important to carry out the research on the national defense R&D project termination decision-making in the theory and practice.The traditional R&D project termination decision-making solutions were divided into two categories:classification-type and regression-type model.On the condition of the analysis on the shortcomings of these two kinds of ways and considering the characteristics of national defense R&D project,a project termination decision-making method based on composite support vector machine(SVM) was proposed,at the same time,a cross-validation approach and particle swarm optimization(PSO) algorithm were adopted to achieve the optimization of the model.Through the establishment of a composite decision-making model,a definite conclusion on the project termination decision-making can be drawn.The application analysis shows that the method can achieve the project classification and sorting,which can make use of more comprehensive information,draw more detailed conclusions and has a good adaptability.