Robust identification of input nonlinear block structure systems

Abstract

A robust identification algorithm is presented for nonlinear systems with disturbance, which is fitted by Hammerstein block structure model. A hierarchical least squares method is proposed to estimate model parameters and track disturbance in combination with auxiliary modelling strategies and separable techniques. The multi-innovation technology for error updating is used to augment the dimension of the innovation matrix, in order to reduce the estimation error variance and enhance the convergence stability of the algorithm. The time-varying disturbance is still tracking by a single innovation strategy. Two adaptive forgetting factors are proposed to enhance the system parameters' convergence characteristics and to improve the track ability of time-varying disturbance. An example is applied to validate the benefits of the proposed algorithm. The established model can facilitate controller design and system operation monitoring.