摘要
运用人工神经网络方法,建立药剂配方、原材料粒度与药剂爆热的三层BP网络模型,通过10组数据对网络进行智能驯化,利用驯化后的网络对另5组数据进行预测,并对这5组数据进行实际测量,结果表明该BP网络能够反映药剂配方、原材料粒度与药剂爆热之间的关系,预测精度较好,相对误差较小。
The artificial neural network model is described. A three-layer BP network model for composition formulation, raw material size and explosive heat was established. Training the network with 10 sets of data, and five other groups of data were predicted using the network. The five sets of data were actually measured. The results the BP network can reflect the relationship between the prescription and the particle size of raw material and the explosive heat. The prediction accuracy is good and the relative error is small.
引文
[1]党建武.神经网络技术及应用[M].北京:中国铁道出版社,2000.
[2]王伟.人工神经网络原理与应用[D].北京:北京航空航天大学出版社,1995.
[3]黄俊,周申范.人工神经能网络法预测炸药爆速的研究[J].火炸药学报,2000(1):34-37.
[4]孙忠华,王珏,瞿中.基于GA的神经网络在火炸药剂辨识中的应用[J].微机发展,2003(8):3-5.
[5]刘记军,唐德高,王春辉,许晓军.结合BP神经网络的遗传算法优选PEE制备参数[J].解放军理工大学学报(自然科学版),2005(4):378-381.
[6]马贵春,张树生,张景林.人工神经网络方法在火炸药科学领域应用进展[J].火工品.2004(1):42-44.
[7]马忠亮,徐方亮,刘海燕,张文才.基于人工神经网络和混合遗传算法的炸药爆速预测[J].含能材料,2007(6):637-640.
[8]韩力群.人工神经网络教程[M].北京:北京邮电大学出版社,2006:14-17.
[9]高隽.人工神经网络原理与仿真应用[M].北京:机械工业出版社,2005:25-35.