Bilinear continuous-time systems identification via Hartley-based modulating functions (Q1295150)

The paper is focused on bilinear continuous-time systems identification using Hartley modulating functions. The structure of bilinear continuous-time systems is thoroughly discussed. In particular, it highlights the applicability of the Hartley modulating functions method for identification of bilinear continuous time systems from recorded input and noise-contaminated output data. The consideration provides an insight into parameter estimation of a wider range of nonlinear systems in practice. In the applied methodology I/O differential equations representing the dynamic system behavior are replaced by the Hartley spectrum equation. As a result it involves the known derivatives of the modulating function instead of the derivatives of the input and noisy output data by applying integral transformations to signals. A frequency weighted least-squares algorithm is used for identification and a normalized root mean-square criterion is employed to make some computational considerations and to investigate the bias of the estimates. The effectiveness of the proposed approach is illustrated by an example. Some discussions and conclusions as well as future research plans are also included in the paper.