Font Size: 

The Bouc-Wen model is commonly used to describe the hysteretic behavior of some mechanical systems. In this model the memory effects of inelastic behavior of the restoring force depends not only the instantaneous displacement but also on the history of the previous displacement. Among different approaches that can be used to describe the hysteretic behavior, the Volterra theory is an interesting strategy, since it represents the response of a system as a sum of linear and nonlinear components using multiple convolutions. However, the Volterra series are not able to describe non-smooth nonlinearities such as in systems with hysteresis. In order to overcome this problem, the present paper propose an adaptive filtering for estimation of the Volterra kernels. The coefficients of the Volterra kernels are identified by using the well known recursive least squares (RLS) filter. To illustrate the performance of the adaptive Volterra model, several amplitude inputs signals was applied in a Bouc-Wen oscillator. The biggest advantage of using a RLS filter combined with Volterra series is the quickly convergence for any input signal. The applicability and drawback of the proposed algorithm are also deeply explored.