Identification of a Parametric Model of the Human Operator in Closed-Loop Control Tasks

Abstract

A novel identification method is implemented to identify on-line a parametric model of the human operator in closed-loop tracking tasks. The method, recently developed by the authors, identifies time delay, form and order, and the minimum essential number of parameters of a continuous differential equation representation. Thus the variance of the estimates is small. Results of exploratory identification tests including stationary, time varying, and secondary tasks are described. The tests with stationary plants demonstrate good convergence and low variance. In the time varying tests, the changing parameters in the human operator were tracked, and the effect of a secondary task loading clearly showed up as significant changes in model parameters. The results demonstrate the applicability of the method to the investigation of human operator adaptive processes in the control of nonstationary control tasks.