An Identification Algorithm That Is Insensitive to Initial Parameter Estimates

This paper is concerned with the estimation of parameters in a constant coefficient, linear system using measurements of the system input and output. Two general methods can be used to estimate these parameters: the equation error method and the output error method. The equation error method is characterized by a single step solution that does not require a prior estimate. Unbiased noise in the output, however, causes a bias in the estimated parameters. The output error method is characterized by iterative solution techniques that require a prior estimate of the unknown parameters. This method provides an unbiased estimate. In this paper, a single estimation procedure is presented that uses the best features of both methods. It does not require a prior estimate of the unknown parameters and unbiased noise in the output will not cause a bias in the final estimate. The method is applied to simulated and flight data.