Synthesis of an Optimal Input to Enhance the Identification of a Linear Multistage System
Rept. for 1 Jul 1971-5 Jan 1972
AEROSPACE CORP EL SEGUNDO CA ENGINEERING SCIENCE OPERATIONS
Pagination or Media Count:
In fitting a transfer function model to inputoutput data, it is sometimes possible to select a particular input with the purpose of minimizing estimation error. In the report, some aspects of this optimal input synthesis problem are examined for a class of linear multistage systems. In particular, an input is selected from a class of equal-energy inputs which maximizes the trace of the Fisher information matrix with respect to the input sensitivity parameters, where it is assumed that the transient response and prediction accuracy of the unforced system is known. The optimal input is compared to a purely random input which has the same average energy, and numerical results for a second-order system are provided. Input synthesis with regard to the other model parameters which determine its unforced behavior is hampered by the fact that the Fisher information matrix is a function of these parameters.
- Statistics and Probability