应用改进的BP神经网络预测导弹的研制费用  

Study of the Application of Improved BP Neural Networks to Development Cost Prediction in Missle

作　　者：梁宇[1] 李保林[1] 李勇[1] 李执力[1] 

机构地区：[1]防空兵指挥学院,郑州450052

出　　处：《战术导弹技术》2006年第2期48-53,共6页Tactical Missile Technology

摘　　要：结合导弹武器系统研制的特点,分析并确定了影响导弹武器系统研制费用的主要因素,并在前人研究的基础上,提出了基于改进的BP神经网络的费用预测模型.该模型不仅克服了以往由于专家主观因素而造成的负面影响,而且提高了算法的运行效率和模型的预测精度.通过具体的实例分析比较,充分证明了利用该方法的科学合理性.最后探讨了利用BP神经网络进行数值预测的方法和应该注意的问题,并给出了一些有益的建议.Combined with the characters of missile weapon system development, the principal factors which affect the development cost are analyzed and defined. Based on the existing research fruits, the model of prediction development cost is presented based on improved BP-NN. The model not only overcomes the negative effect of subjectivity from exports, but also improves the process of the program and its prediction precision. The practical calculation proves the method is scientific. At the same time, the method of how to use BP-NN in numerical prediction and the problems which should be paid attention to are discussed, and some helpful advice is given.

关 键 词：BP神经网络 改进算法 研制费用 预测 

分 类 号：TJ760[兵器科学与技术—武器系统与运用工程]