NEURAL-NETWORK FOR STRUCTURAL DYNAMIC-MODEL IDENTIFICATION

The identification and modeling of linear and nonlinear dynamic system s through the use of measured experimental data is a problem of consid erable importance in engineering. Among the identification methods, th e artificial neural network is a newly developed technique. Due to its attributes, such as parallelism, adaptability, robustness, and the in herent ability to handle nonlinearity, artificial neural networks have shown great promise in function mapping, pattern recognition, image p rocessing, and so on. However, dynamic function mapping, including the structural dynamic model identification, is still a challenging topic in neural network applications. A neural network approach for structu ral dynamic model identification is presented in this paper. The neura l network is trained, tested, and verified by using the responses reco rded in a real apartment building during earthquakes. The results show that the dynamic behaviors of the building can be very well modeled b y the trained neural network. The results also indicate the great pote ntial of using neural networks in structural dynamic model identificat ion.