Identification of human response models in manual control systems
A. V. Balakrishnan, Lawrence W. Taylor · NASA Technical Reports Server (NASA) · 1967
Frequency domain and time domain methods of analysis are reviewed with regard to their application toward identifying pilot models.The models would subsequently be used to study the stability and performance of a man-machine system in which the human controller performs a compensatory tracking task.Sample linear model results are corLpared and discussed.The inherent requirement constraining the freedom of the form of the pilot model is also discussed.The constraint in the frequency domain consists of smoothing with respect to frequency; whereas, the constraint for the time domain model is more natural and meaningful in that it consists simply of limiting the memory of the pilot model.The linear models determined by both methods were almost identical.The time domain method of analysis enables the determination of a nonlinear pilot model.The inclusion of a cubic as well as a linear term accounted for only a small additional part of the pilot's remnant and indicated that only a small portion of the total power of the pilot's output is caused by nonlinearities.The power spectral density of an ensemble average of the pilot's outputs is used to determine the upper limit on the amount of power associated with a deterministic response.The indication is that about half the remnant is stochastic and only a small part is due to nonlinear and timevarying response for the example discussed.