An Alternative Forcing Function for Identifying Pilot Frequency Response Using a Series of Mesas

The command inputs selected for system identification (SYSID) are dictated by numerous factors, some of which include: 1) The frequency range of interest; 2) The capability of the system to sustain the inputs; 3) The capability of the system to remain ‘agnostic’ to future inputs. When the elements comprising, the system being identified are all electro-mechanical, frequency sweeps, sum-of-sines, and impulsive inputs are standard identification techniques. However, when human manual control becomes an element of the system, the second and third factors are key considerations. Sum-of-sines (SOS) has been used extensively for identifying human control dynamics as it provides an input that is perceived by the pilot as random and focuses power at discrete frequencies. A disadvantage of SOS is the attentional demand it requires from the human operator, which limits the duration of an identification run to typically around one minute. This in turn constrains the lowest frequencies that can be identified, and multiple consecutive runs can lead to operator fatigue and performance degradation. Discrete inputs such as ramps have been employed with human-in-loop testing, but only with regard to Handling Qualities and performance testing. This work examines discrete inputs as a method for human-in-loop SYSID. An experiment was conducted using two terrain profiles: 1) Pop-up (POP), where four mesas (hills) of varying height, slope, and plateau length were unevenly spaced on flat terrain; 2) Sum-of-sines (SOS), where the terrain was comprised of the sum of eleven non-harmonically-related sine waves, so that the contour was perceived as varying randomly. The task for both terrain types was to maintain 20 feet above the ground using pitch (airspeed was fixed at 35 knots) for each run. Bedford workload ratings were collected after each run. Both terrain profiles produced approximately the same open-loop frequency responses, and their coherences were not significantly different. However, the Bedford ratings showed the POP profile was significantly easier to execute than SOS. The POP technique thus presents a less demanding, more appealing, and potentially more consistent way for eliciting frequency information relating to pilot gain, stability, and time delay.