Redesign of a Variable-Gain Output Feedback Longitudinal Controller Flown on the High-Alpha Research Vehicle (HARV)

SummaryThis paper describes a redesigned longitudinal controller that flew on the High-Alpha ResearchVehicle (HARV) during calendar years (CY) 1995 and 1996. Linear models are developed for both themodified controller and a baseline controller that was flown in CY 1994. The modified controller wasdeveloped with three gain sets for flight evaluation, and several linear analysis results are shown com-paring the gain sets. A Neal-Smith flying qualities analysis shows that performance for the low- andmedium-gain sets is near the level 1 boundary, depending upon the bandwidth assumed, whereas thehigh-gain set indicates a sensitivity problem. A newly developed high-alpha Bode envelope criterionindicates that the control system gains may be slightly high, even for the low-gain set. A large motion-base simulator in the United Kingdom was used to evaluate the various controllers. Desiredperformance, which appeared to be satisfactory for flight, was generally met with both the low- andmedium-gain sets. Both the high-gain set and the baseline controller were very sensitive, and it was easyto generate pilot-induced oscillation (PIO) in some of the target-tracking maneuvers. Flight target-tracking results varied from level 1 to level 3 and from no sensitivity to PIO. These results were relatedto pilot technique and whether actuator rate saturation was encountered.IntroductionDuring spring and summer calendar year (CY) 1994, two NASA research flight controllers wereflown on the High-Alpha Research Vehicle (HARV). To maximize flight experiment time, one control-ler was designed for the longitudinal axis and the second controller for the lateral-directional axes. Thelongitudinal controller methodology is called variable-gain output feedback (refs. 1 through 4) and isthe controller that is discussed in this paper. The lateral-directional methodology (ref. 4) is calledCRAFT and was combined with a pseudo controls Mending approach.The Variable-Gain methodology (ref. 1) was developed to overcome shortcomings of the traditionalapproach to gain scheduling. In the traditional approach, constant-gain feedback control laws aredesigned individually at many operating points over the flight envelope. These feedback gains are thencombined by using a curve-fit technique (interpolation, straight-line approximation, or a least squaresfit) to generate a gain schedule for the final control gains. Several schedules are often used and com-bined when more than one independent variable is involved. Depending upon the curve-fit techniqueused, stability could become an issue.Variable gain is an integrated design approach in which all design operating conditions are handledsimultaneously, creating a more efficient design process. A gain functional is optimally produced withinthe design algorithm and consists of measured scaling parameters selected by the designer and associ-ated designed gain-matrix components. Feedback gains are calculated continuously during flight, result-ing in a smooth-gain schedule. The control system is guaranteed to be stable at all design operatingpoints. Other features of the methodology include (1) output feedback that allows all dynamics to beincluded in the design process, (2) optimal control that allows tradeoff between states and controls,(3) stochastic design that allows process noise and sensor noise to be included, and (4) direct digitaldesign for applicability to digital computers. Thirty-nine design flight conditions were used for theflight controller described in this paper (ref. 3).The HARV provided a test bed to demonstrate high-alpha control methodologies using advancedcontrol effectors. In particular, thrust vectoring was the technology used to allow maneuvering duringpost-stall. The thrust-vectored controls were installed on a modified F/A-18 airplane and deflected thethrust to provide pitch and yaw motions. Research on this airplane was part of the High-Alpha Technol-ogy Program (HATP) (ref. 5) that included other high-alpha related experiments in technology areassuch as propulsion, aerodynamics, loads, sensors, and other advanced control effectors.There were several guidelines that influenced the control design. Pitch agility was one of the mostimportant longitudinal high-alpha design guidelines. An example pitch-up agility guideline is the