基于神经网络灰色Verhulst算法的装备研制费用预测模型  

Forecasting model for development cost of equipment based on artificial neural network and grey Verhulst algorithm

作　　者：刘建永[1] 李凌[1] 伍中军[1] 陈玉金[1] 

机构地区：[1]解放军理工大学工程兵工程学院,江苏南京210007

出　　处：《解放军理工大学学报（自然科学版）》2008年第4期335-338,共4页Journal of PLA University of Science and Technology(Natural Science Edition)

基　　金：国家973计划资助项目(2007CB411805);国家自然科学基金资助项目(40405010)

摘　　要：为了提高装备寿命周期费用LCC(life cycle cost)中研制费用的预测精度,对于呈S型的研制费用累加序列,利用灰色Verhulst模型对其预测。构造基于时间序列的人工神经网络输入输出模式对,利用BP神经网络对灰色Verhulst预测结果进行训练。仿真实例表明,该组合算法预测结果比单纯使用灰色Verhulst模型所得到的结果总体误差要小,将神经网络引入到灰色Verhulst模型中能较好地提高预测精度。In order to enhance the forecast precision of development cost in equipment life cycle cost （LCC）, grey Verhulst model was applied to forecasting development cost accumulation sequence with the shape of alphabet S. Couples of input and output data of artificial neural network based on time sequence were formed, and BP neural network utilized to train the grey forecast result. The simulation example shows that the hybrid algorithm has fewer errors than grey Verhulst model. The introduction of BP net-work into grey Verhulst model advances the forecast precision.

关 键 词：装备 灰色VERHULST模型 人工神经网络 研制费用 时间序列 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] E919[自动化与计算机技术—计算机科学与技术]