人工神经网络在风机振动控制.pdfVIP

Proceedings of the 24th Nordic Seminar on Computational Mechanics J. Freund and R. Kouhia (Eds.) c Aalto University, 2011 Artificial Neural Networks for Nonlinear Dynamic Response Simulation in Mechanical Systems 1,2 1 3 Niels H. Christiansen , Jan Høgsberg , and Ole Winther (1)Department of Mechanical Engineering, Technical University of Denmark, Kgs. Lyngby, Den- mark, e-mail: Niels.Christiansen@ (2)DNV Denmark, Hellerup, Denmark (3)Department of Informatics and Mathematical Modeling, Technical University of Denmark, Kgs. Lyngby, Denmark Summary. It is shown how artificial neural networks can be trained to predict dynamic response of a simple nonlinear structure. Data generated using a nonlinear finite element model of a simplified wind turbine is used to train a one layer artificial neural network. When trained properly the network is able to perform accurate response prediction much faster than the corresponding finite element model. Initial result indicate a reduction in cpu time by two orders of magnitude. Key words: Nonlinear structural dynamics, Artificial neural networks. Introduction Time domain simulation of nonlinear systems using finite element method (FEM) analysis can be computationally very expensive - especially in fatigue calculations where long response his- tories are needed in order to obtain reliable time series statistics. The use of artificial neural networks (ANN) combined with classical methods have shown promising results in reducing this computational cost [1]. This paper presents a hybrid method for simulation of dynamic response of a simple nonlinear structure. It is shown how an ANN can be trained to predict dynamic response of a simplified model of a wind turbine. FEM models of nonlinear structures often require fine element mesh discretization, small time step size and ite