神经网络在点火药性能预测中的应用

Application of Neural Network in Prediction of Ignition Powder Performance

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.