基于挣值分析和风险管理，通过蒙特卡洛模拟获取项目数据，使用二次判别分析、随机森林和支持向量机进行模型学习和完工预测是项目控制的有效方法之一。在现有研究基础上，考虑项目执行过程中的剩余工作时间、剩余工作费用和风险，分别应用现有研究方法、梯度提升树和人工神经网络进行模型学习，利用嵌套交叉验证进行模型选择和模型评估。研究结果表明，优化后的方法显著提升项目完工预测的准确率。

Based on the analysis of the earned value and the risk management, the project data is acquired through the Monte Carlo simulation, and the model learning and the completion prediction by using the quadratic discriminant analysis, the random forest and the support vector machine are one of the effective methods of the project control. On the basis of existing research, this paper takes into account the residual working time, the remaining work cost and the risk in the execution of the project, and applies the existing research methods, the gradient lifting tree and the artificial neural network to study the model, and makes model selection and model evaluation by using the nested cross validation. The results show that the optimized method can improve the accuracy of project completion prediction.

C935;F224;G301

军事科学院科研项目“项目管理相关概念及案例研究”（Y85301X1G4）