System Identification of Non-Linear System Using Gentic Algorithm for Development of Active Vibration Control Algorithm

Abstract

This work presented parametric linear estimation techniques for dynamic modelling of an extremely nonlinear flexible plate system for the development of active vibration control. In this research, a simulation study using Finite Difference method (FDM) was developed to create the vibration data of flexible rectangular plate structure with all edges clamped (C-C-C-C) boundary conditions. A finite duration step input force was then applied to excite the system at excitation point and the dynamic responses of the system at detection and observation points were investigated. An autoregressive with exogenous inputs (ARX) model with the order of 10 was
proposed as a suitable model and the corresponding parameters were predicted using Genetic Algorithm (GA). The algorithm attained a suitable mean square error of 0.00028 in the 117th generation. Finally, the validity of the obtained model was investigated using prediction and statistical measurements. The parametric identification algorithm thus developed forms a suitable platform for the development of an active vibration control strategy for vibration suppression in the future work.