基于神经网络模型的动载荷识别

依据结构动力学理论推导了在时域中用于神经网络算法的自回归函数，相应建立了具有时延反馈的神经网络动载荷识别模型。阐明了这种网络的基本学习算法和回忆算法。数值仿真和试验件的验证试验表明该神经网络模型用于动载荷识别时具有精度高、无累积误差、抗干扰能力强等优点，并且适用于各种类型的动载荷，尤其对冲击载荷的识别更具有独特的优势。该模型在动标学习过程中要求信息量小，试验成本低，是一种非常值得在工程中推广应用的新型动载荷识别方法。

The Dynamic Load Identification Research Based on Neural Network Model

In this paper,a recursive function for neural network computation in time domain is derived by means of structural dynamics and then the neural network model for dynamic load identification with time delay is established.The learning and recall processes of the network are presented.It is shown by numerical simulations and experiments that the network model has many advantages,such as high accuracy,no accumulation error,robustness to noise and so on,for dynamic load identification.And it can be used to identify various kinds of dynamic load,especially for the transient load.The neural network model has low cost and requires less information,so that it is a very useful dynamic load identification method in engineering.