航空学报 > 2010, Vol. 31 Issue (10): 1907-1913

过载与弹射速度关系研究及神经网络实现

毛晓东, 林贵平, 郁嘉   

  1. 北京航空航天大学 航空科学与工程学院
  • 收稿日期:2009-11-23 修回日期:2010-04-30 出版日期:2010-10-25 发布日期:2010-10-25
  • 通讯作者: 林贵平

Study of Relationship Between Load Factor and Ejection Velocity and Its Neural Network Implementation

Mao Xiaodong, Lin Guiping, Yu Jia   

  1. School of Aeronautic Science and Engineering, Beijing University of Aeronautics and Astronautics
  • Received:2009-11-23 Revised:2010-04-30 Online:2010-10-25 Published:2010-10-25
  • Contact: Lin Guiping

摘要: 弹射速度是弹射座椅双态程序控制的主要输入参数之一,其和弹射高度一同决定了救生伞的开伞时间。试验发现,在某些特定的条件下,弹射速度的测量会出现较大误差从而严重影响弹射救生系统的救生性能。提出一种根据弹射座椅出舱瞬间人椅系统体轴 x 方向过载( nx )值判断弹射速度的方法。建立了人椅系统出舱阶段的数学模型,在MSC.EASY5基础平台上开发了模块化的仿真模型,并基于批处理原理进行了求解器设计。通过数值仿真,建立了平飞状态不同弹射高度及弹射离机质量下体轴 x 方向过载值与弹射速度之间的关系曲线。利用基于误差反向传播算法的人工神经网络,即BP神经网络实现了输入向量(弹射高度及体轴 x 方向过载)到输出值(弹射速度)之间的连续非线性映射。分析了不利姿态参数对关系曲线的影响,在满足工程要求的情况下,可以忽略其影响。用本文方法判断得到的弹射速度与地面弹射试验数据进行了比较,结果表明误差满足工程要求,可以作为弹射速度测量的一种余度设计。

关键词: 过载, 弹射速度, 惯性测量, 仿真, 神经网络, 不利姿态

Abstract: The ejection velocity of an ejection seat is a primary input parameter of the dual mode sequencer ope-ration which, along with the ejection altitude, determines the parachute-opening time. It is found that in certain circumstances large errors of the sensed velocity may exist which significantly influence the performance of the escape system. In this study, a method that predicts the ejection velocity by using load factor ( nx ) in the x direction of the body-axis system is presented. A mathematical model of the egress phase is established, and its module simulation model is set up on the MSC.EASY5 fundamental platform. Furthermore, a solver based on the batch program is designed. According to numerical simulation, the relationship curves between the nx and ejection velocity at various altitudes and gravities are obtained. Subsequently, back propagation (BP) neural networks are established in order to implement the sequential nonlinear mapping from input vectors (altitude and nx ) to the ejection velocity. The impact of adverse attitude parameters on the relationship is analyzed, which is found to be negligible under engineering requirements. Finally, the predicted ejection velocity is compared with the experimental data and the error is acceptable for engineering applications.

Key words: load factor, ejection velocity, inertia measurement, simulation, neural network, adverse attitude

中图分类号: