| Summary: | 碩士 === 國立成功大學 === 航空太空工程學系 === 86 === The purpose of the present study is to identify projectile's
weight and its associated impact velocity utilizing neural
network scheme based on strain gauges measured stress wave
information. A series of tests were conducted in order to
measure the stress wave propagated in the target rod. Circular
steel rod with a diameter of 12.7 mm and ranging from 20 cm to
40 cm were used as the projectile in the experiment. Target rod,
with a length of 100cm, made of steel was struck by the impact
rod at impact velocities ranging from 10 m/s to 20 m/s. A total
of 130 impact tests with five different projectile length
traveling at various incident velocity were conducted in this
work. Strain gauges(Tokyo Sokki Kenkyujo Co., Ltd., FLA-5-11-1L)
were mounted in the target rod in order to measure the stress
wave signal. Fifteen measured time domain impact signals are
used as the input data and the corresponding projectile's
weight and its impact velocity are regarded as the output data
in the neural networks. The time-domain back-propagation neural
network with adaptive learning rate scheme is adopted in the
present work. The trained network is capable of identifying the
projectile's weight and its impact velocity for untrained cases
closely.
|
|---|
