Statistical Identification of a Human-Operator Model in Control Systems Subject to Random Disturbances,
Abstract
In investigating human-operator factors in manual-control systems subject to random disturbance effects, one can accept the hypothesis of a quasi-linear model assuming linearity and stationariness. For this purpose, one should identify the parameters of such a model, assuming that the operator acts, basically, as a linear member with a transmittance H(jw) to a linear object G(jw), with the forcing action u(t) a stationary and ergodic random process. In order to find the optimum operator performance (in terms of a rms linear approximation of H(jw)), one has to minimize the functional J = E ((c(t) - c'(t)) squared), with c(t) the real reaction measured at the operator's output and c'(t) the desired model reaction. Solution of the problem is best found in the frequencies region. A study was made of several different control situations which occur most frequently. Two specially-trained operators were tested under varying operating conditions. It was found that the adopted method proved well-chosen in all cases where hypothesis of a quasi-linear model for the statistical identification of dynamic human-operator factors is acceptable, and the action of the operator involved may be compared with the action of a 'classical' continuous regulator.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 08, 1968
- Accession Number
- AD0849370
Entities
People
- A. P. Firla
Organizations
- National Air and Space Intelligence Center