R68-6 Regression Analysis and Parameter Identification

This paper was written in reaction to a body of literature on continious parameter tracking that has by and large ignored the use of continuous regression techniques introduced by A. I. Rubin in an earlier paper.¹The case for the use of continuous regression in the parameter tracking problem is presented in a concise and professional manner. A distinction is made between t, the "real world time" employed in the formulation of the model, and in the development of the regression performance measure, and τ, the "data processing time" employed in the steepest descent circuits that are performing the minimization of the performance measure with respect to the parameters. This distinction is an important contribution, both analytically and conceptually, to the parameter identification literature; the authors\´ belief that other workers in the field might profit from this distinction is well founded.