Performance assessment for a class of nonlinear systems

Performance assessment for nonlinear control systems is of great current interest. In this paper, a class of nonlinear systems are concerned. The analytical benchmark performance and the associated benchmark controller are established by minimizing the variance of the measurement output. Similar to filtering and correlation analysis (FCOR) algorithm, the calculation of performance index in higher-order statistics forms is proposed. To facilitate assessment, a filtering or whitening method is also presented. The white noise sequence is estimated by dynamic principal component analysis (DPCA) based on the nonlinear autoregressive exogenous (NARX) model. By comparison with traditional method, the proposed method can effectively avoid biased estimation of control performance for nonlinear systems. The numerical example indicates the effectiveness of the proposed performance assessment method.