Analysis of Gain Scheduling for the Three-Loop Autopilot Structure

A new approach to gain scheduling the classical \three-loop" autopilot structure is discussed for both the single-axis (pitch) case and the three-axis case. Simulation results in the single-axis case show that the new gain scheduled controller yields better performance than a traditional approach. Analysis of the three-axis case indicates an even greater potential for performance improvement with little or no increase in controller complexity over a traditional approach. 1 Single-Axis Analysis We rst consider gain scheduling issues for autopilot (controller) design for a pitch-axis missile model (plant). The plant input is commanded tail deeection (t), the output is normal acceleration (t), and measured state variables are pitch rate q(t) and angle of attack (t). Direct controller inputs are commanded normal acceleration c (t), normal acceleration, and pitch rate. Scheduling-variable inputs to the controller are angle of attack and the exogenous variables Mach number M(t) and altitude h(t). These notations are summarized in Figure 1. The objectives for the closed-loop system are that (t) track c (t) with asymptotically zero error, with less than 5% overshoot to a step command, and that various other performance speciications, including stability and robustness speciications, be satissed while M(t) and h(t) vary over a range of values. Controller Plant-? ? 6 6 M(t); h(t) c(t) (t) (t) (t); q(t) Figure 1: The plant and controller structure.