ANALYSIS OF VOLTERRA/WIENER NEURAL NETWORKS FOR ADAPTIVE IDENTIFICATION OF HYSTERETIC SYSTEMS

Abstract eng:
This study attempts to demystify a powerful neural network approach, Volterra/Wiener Neural Network (VWNN), for modeling nonlinear hysteretic systems and in turn to streamline its architecture to achieve better computational efficiency. Artificial neural networks are often treated as ”black box” modeling tools, in contrast, here the authors examine problem formulation and network architecture to explore the inner workings of this neural network. Based on the understanding of the dynamics of hysteretic systems, some simplifications and modifications are made to the original VWNN in predicting accelerations of hysteretic systems under arbitrary force excitations in an off-line or even in an adaptive (on-line) mode. The VWNN is able to yield a unique set of weights when the values of the controlling design parameters are fixed. One training example is presented to illustrate the application of the VWNN.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket