摘要
针对推进剂中细高氯酸铵含量及粒度对推进剂力学性能、燃速的影响,分别建立广义回归神经网络(GRNN)与遗传算法反向传播(GABP)神经网络模型。建模所用网络训练数据是推进剂配方中不同粒度的高氯酸铵含量对应的不同温度下测试的推进剂的抗拉强度、断裂伸长率及燃速共8组数据,而用第9组数据的细高氯酸铵的粒度及含量作为模型输入,对不同温度下测试的推进剂的抗拉强度、断裂伸长率及燃速进行预测。结果表明,2种神经网络模型的预测值与实验值均具有较好的吻合性,其最小相对误差分别为0.19%、0.45%。所建模型对配方中细高氯酸铵的级配调整具有指导意义。
Aiming at the influence of the content and particle size of fine ammonium perchlorate in the propellant on the mechanical performance and burning rate of the propellant,the general regression neural network (GRNN) and genetic algorithm back propagation (GABP) neural network models were established respectively.The network training data used in the modeling is 8 groups of data altogether of the tensile strength,elongation at break and burning rate of the propellant tested at different temperatures corresponding to the content of ammonium perchlorate with different particle sizes in the propellant formulation.The particle size and content of fine ammonium perchlorate in the No.9 group of data are used as the model input to predict the tensile strength,elongation at break and burning rate of the propellant tested at different temperatures.The results show that the predicted values of the two neural network models agree well with the experimental values,their minimum relative errors are 0.19% and 0.45%,respectively.The established models have guiding significance for the gradation adjustment of fine ammonium perchlorate in the formulation.
作者
张高章
刘晶晶
Zhang Gaozhang;Liu Jingjing(Jiangxi Aerospace Jingwei Chemical Co.,Ltd.,Ji’an 343711,China)
出处
《化学推进剂与高分子材料》
CAS
2021年第2期61-64,75,共5页
Chemical Propellants & Polymeric Materials
关键词
广义回归神经网络
推进剂
遗传算法反向传播神经网络
预测
general regression neural network
propellant
genetic algorithm back propagation neural network
prediction