Mechanical properties prediction of HTPB propellants based on machine learning method
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
维普
万方数据
固体推进剂中AP颗粒的级配不同会导致力学性能相差巨大,为了探究AP颗粒的粒度值和质量分数对端羟基聚丁二烯(HTPB)推进剂力学性能的影响规律,采用机器学习的方法对推进剂的力学性能进行仿真预测,降低了实验成本并提高了预测效率.首先,对不同级配的HTPB推进剂进行拉伸试验,得到不同温度下抗拉强度和伸长率;其次,以拉伸试验结果为样本进行机器学习,分别构建了反向传播(Back Propagation,BP)神经网络、粒子群算法优化的反向传播(Particle Swarm Optimization Back Propagation,PSOBP)神经网络和遗传算法优化的反向传播(Genetic Algorithms Back Propagation,GABP)神经网络对推进剂的力学性能进行预测.结果表明,力学性能与颗粒级配的内在关系较为复杂,并非简单的线性关系.PSOBP和GABP可以用于预测不同级配下HTPB推进剂力学性能,并且GABP神经网络可以更好地预测推进剂的力学性能变化.
Different particle size distributions of AP particles in solid propellants can result in significant differences in mechan-ical properties.In order to investigate the influence of particle size and mass fraction of AP particles on the mechanical properties of hydroxyl-terminated polybutadiene(HTPB)propellants,the machine learning method is used to predict the mechanical properties of HTPB propellants,reducing experimental costs and improving prediction efficiency.Firstly,tensile tests were conducted on HTPB propellants with different particle size distributions to obtain the tensile strength and elongation at different temperatures.Secondly,machine learning was performed using the results of the tensile tests as samples,and three neural networks,namely back propagation(BP)neural network,particle swarm optimization back propagation(PSOBP)neural network,and genetic algorithms back propaga-tion(GABP)neural network,were constructed to predict the mechanical properties of the propellants.The results indicate that the re-lationship between mechanical properties and particle size distribution is complex and not simply linear.PSOBP and GABP can be used to predict the mechanical properties of HTPB propellants under different gradations,and the GABP neural network performs better in predicting variations in the mechanical properties of the propellants.
维普
万方数据
