Minimal representations of human operator dynamics in compensatory systems

摘要

Classical operator models of fixed forms have been used to successfully reproduce experimental frequency response data over various frequency bands. ^Optimal control operator models (OCMs) have also successfully duplicated experimentally obtained frequency response data over a fairly wide frequency range. ^It is shown that, for single axis compensatory tracking tasks, use of frequency-weighted variables in the OCM performance index allows an OCM method to better match experimental data over the entire frequency range for which data is typically available; unique descriptions of the human operator are not limited to the analysis of single-degree-of-freedom tracking tasks. ^Projection methods have recently been developed to eliminate complexity and over-parameterization associated with OCM based methods and have resulted in the fixed order modified optimal control model (FOMOCM). ^A precision classical model and FOMOCMs of varying complexity are compared to experimental frequency response data for a simple divergence controlled element to determine the minimal order model that maintains the salient features of the experimental data. ^Application of the FOMOCM algorithm to other controlled elements supports the conclusion that a minimal representation of the human operator in a single axis compensatory tracking task is a fifth order transfer function plus a time delay. ^(Author)