振动环境试验响应的神经网络预测方法

NEURAL NETWORK PREDICTION FOR RESPONSES OF RANDOM VIBRATION ENVIRONMENT TESTS

航空、航天、机械等领域中的大型复杂结构的地面振动环境试验是检验产品或结构对振动环境的适应性以及评估动态性能的重要手段之一。然而,由于试验成本高和试验中测点布置难度和测点数量有限,只能反映结构在有限的试验条件下局部测点位置上的响应情况。本文通过神经网络方法建立结构测量点的响应和输入的映射关系,可以通过较容易的振动环境试验结果,对所建立的神经网络模型进行训练学习以识别映射关系模型,从而对结构在较严酷的环境条件下的试验结果进行预测。本文通过一典型结构的振动环境试验以及该方法的实现证明了神经网络应用于振动环境试验的响应预测方法的有效性和可行性。

Vibration test is one of the very importnt environmental tests used for calibrating large and complex products in engineering fields,such as,aeronautics,astronautics and mechanical engineering.It is often difficult to complete a perfect vibration survey with several limited vibration tests.An artificial neural network method is developed for modeling the input-output measurement data obtained from a random vibration test.The established neural network model is used to predict the responses of another random vibration tests.Two random vibration environment tests of a combined structure are investigated.The power spectral density(PSD) and the root mean square(RMS) for the vibration responses of the combined structure are analyzed from both the predicted results and the measured results.The results show that the proposed method is feasible and effective.This method is particularly applicable to large and complex structures under random vibration environment test.