Hybrid Modeling for Structural Dynamics of Complex Systems Based on Neural Network

摘要

To vibration isolation system composed of four wire ropes for the study, and a hybrid modeling method was adopted which neural network mode and a priori knowledge model are connected in series. Multi-DOF hybrid neural network model is established. A method of determining indirectly restoring force of isolation system by the experimental data was proposed through the establishment of the rigid body dynamics equation of vibration isolation system. Through the network training, the neural network model to describe the restoring force characteristics of the four wire rope isolator was established. The simulation results show that the hybrid neural network modeling method has the advantages of high precision, small amount of the training sample data and the modeling cycle is short, etc. Hybrid model contains the physical parameters of the system, and it is very convenient to analyze the effect of parameters.