导弹飞行过程中动态载荷识别方法

导弹在飞行过程中的外部载荷会导致结构振动,从而引发燃烧不稳定。为了更好地理解燃烧不稳定性,预测外部载荷是很重要的。该文提出了一种基于神经网络的动态载荷识别方法。首先,设计并加工了导弹模型,进行有限元模态分析与模态试验,提取典型模态振动频率,验证了所使用有限元模型的合理性。该方法基于有限元模型开展仿真计算,神经网络由仿真结果中的足够样本数据训练,从而使神经网络能够识别典型载荷。搭建地面试验系统,开展地面激励实验获取相关数据,首先进行典型载荷识别,相对误差可达1.21%,验证了所训练神经网络的准确性和试验系统的可行性。随后对随机载荷进行识别,结果表明,用所提方法识别随机动载荷相对误差小于1.82%,该载荷识别方法具有良好的识别能力。对于导弹结构设计和发动机燃烧不稳定的预测具有重要意义。

Dynamic Load Identification Method During Missile Flight

The external load of the missile during the flight can cause the structure vibration,which may trigger the onset of combustion instability. To better understand the combustion instability,it is important to predict the external load. A dynamic load recognition method based on the neural network was proposed. The missile model was designed,and the finite element modal analysis and modal experiment were carried out,and the typical modal vibration-frequency was extracted. The rationality of the finite element method was verified. The neural network is trained by enough sample data from the simulation results,so that the neural network can recognize typical loads. The relevant experimental modal was manufactured,and the ground test system was set up. The typical load was first identified. The relative error can reach 1.21%,which verifies the accuracy of the trained neural network and the feasibility of the test system. Subsequently,the random load was identified. The results show that the relative error of identifying random dynamic loads is less than 1.82% by the proposed method,and the load identification method has good recognition ability. It is of great significance for the prediction of rocket structural design and engine combustion instability.