NONLINEAR DATA DEPENDENT SYSTEM FOR NONLINEAR RANDOM VIBRATION.

This paper presents a computational method for the identification of nonlinear dynamic systems through random vibration data. The discrete model used is referred to as the nonlinear autoregressive moving average (NLARMA) model which is a direct extension of the conventional time series model. The identification is carried out by a Nonlinear Data Dependent System (NLDDS) modeling strategy. The NLDDS method utilizes a statistical data classification system for nonlinear pattern recognition, and a model search procedure for parameter estimation. It is shown that the NLDDS approach leads to a satisfactory result. The method is applicable to many engineering systems. The paper concludes with an example of the nonlinear chatter process and a discussion of issues associated with this type of modeling.