固体推进剂力学性能预估研究  

Research on Mechanical Property Prediction of Solid Propellant Base on GA-BP Neural Network

作　　者：李进贤[1] 莫文宾[1] 唐金兰[1] 

机构地区：[1]西北工业大学航天学院,陕西西安710072

出　　处：《计算机仿真》2011年第1期76-79,共4页Computer Simulation

摘　　要：固体火箭发动机中,药柱的结构完整性直接关系到发动机的结构完整性和可靠性,而推进剂的力学性能对保持药柱结构完整性起着重要作用,也是决定推进剂寿命的重要指标。为了预估固体推进剂的力学性能,提高系统的可靠性,将遗传算法和神经网络相结合,建立了预估固体推进剂力学性能的遗传神经网络(GA-BP)模型。利用模型预测了某固体推进剂在不同温度、湿度和时间下的抗拉强度、延伸率、弹性模量变化情况,并与试验结果进行了比较。结果表明,模型预估精度高,泛化能力强,仿真计算与试验在结果上有很好的一致性。从而为固体火箭发动机的结构完整性研究提供可靠依据。Solid propellant grain structural integrity influences the structural integrity and reliability of solid rocket motor （SRM）. Mechanical property of solid propellant plays an important role in grain structural integrity, which is critical criterion of solid propellant life. In order to predict mechanical property of solid propellant, a new mechanical property prediction model for solid propellant was established by means of combination of genetic algorithm with neural network （GA-BP）. Using above model, the mechanical property of a solid propellant in conditions of different tem- perature, humidity and time was predicted and compared with experiment results. The comparison results show high precision of the model and strong ability of generalization and with good consistency between prediction of model and experiment. The investigation provides reliable assistance for structural integrity research of SRM.

关 键 词：固体推进剂 性能预测 神经网络 遗传算法 

分 类 号：V512[航空宇航科学与技术—航空宇航推进理论与工程]