AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
This study details the development of a sub-optimal pilot model that blended the classical and optimal pilot model approaches in an attempt to achieve the advantages of each. This model used a numerical solution to the linear quadratic Gaussian problem to find the pilot gain, lead, and lag values that minimized a performance index consisting of task error and control usage. This development was conducted in four phases. First, an optimal pilot model developed by Systems Technology, Incorporated, was analyzed in detail. This analysis included a step-by-step example problem to clarify the models logic and an in-depth sensitivity analysis of the models parameters. Second, a ground and airborne evaluation of human pilot response was conducted using the Calspan variable stability Lear II aircraft. Primary pilot response parameters were recorded and examined using statistical and Fourier transform analysis in an attempt to provide insight into human pilot response. Third, a numerical solution to the linear quadratic Gaussian control problem that allows the compensator form to be predetermined was derived. Finally, the sub-optimal pilot model was developed and an analysis of the models parameters was conducted.