国产飞机研制费用建模因样本容量有限而导致了传统多元统计回归模型的适用性问题。应用支持向量机（SVM）理论进行多维空间小样本建模与估算，以国产运输机研制费为例，基于设计专业分解研制费用样本，采用变量投影重要性指标（VIP）筛选自变量，建立研制费用估算模型，并使用交叉检验与误差概率估计验证模型有效性。该模型方法适用于国产飞机研制费用估算，且可用于飞机改型及改装的经济性定量评估，便于项目早期决策使用。%Because of the lack of samples of china aircraft in modeling,the traditional multivariate regression aircraft cost estimation model can be only used in restricted area. To solve this problem, Support Vector Machine (SVM)theory is applied to deal with the multidimensional space with small sample modeling and estimation. Based on the cost breakdown structure of the specialties according to the cost sample data,the domestic transporter aircraft R&D cost estimation model is built by the combination of Variable Importance in Projection (VIP) for model independent variable selection , cross-validation ,estimation error probability estimates for checking model validity and SVM theory for modeling .The methods of cost modeling discussed here which can estimate the domestic aircraft R&D cost efficiently and reliably,can be easily used for economy evaluation and decision-making of aircraft retrofit in the early stages.
展开▼
