| 초록 |
We propose a continuous-time prediction error method to identify combined deterministic-stochastic continuous-time process. It minimizes the prediction error using the Newton optimization method with analytical derivatives of the objective function with respect to the adjustable parameters. Compared with previous discrete-time identification methods, the proposed method does not suffer from a small sampling period problem. Also, while previous continuous-time approaches using transforms cannot treat a large sampling period, the proposed method can incorporate directly both small and large sampling periods as well as irregular sampling. Unbiased estimates are guaranteed from the proposed prediction error methods for open-loop as well as closed-loop test data sets like discrete-time prediction error method. |
