The current pressure for fuel burn savings and increasing performance in the commercial aerospace market demands highly complex engine control systems to optimize fuel consumption throughout the engine operating envelope, as well as meet the regulatory requirements in terms of safety and performance. These conflicting objectives normally lead to trade-off solutions that are difficult to precisely estimate. Therefore some decisions to characterize the engine controller still reside on experience from previous designs and, as a result, add subjectivity and increase the potential for wrong parameter selection.
This paper proposes an algorithmic approach to design a turbojet engine controller in a multivariable, two-degree-of-freedom configuration, obtaining H-infinity robust stabilization. It introduces an optimized loop shaping design procedure, with the use of a Genetic Algorithm (GA), to further improve the control system performance, as well as bring the experience applied by controller designers and engineers to an automated process, when setting the parameters to shape the frequency response of the engine control loops. The resulting controller is evaluated by computer simulations under typical operating conditions and it is compared to other strategies like a discrete-time Linear Quadratic Regulator with Integral Action (LQI) as well as a Linear Quadratic Gaussian (LQG) controller with Loop Transfer Recovery (LTR).
H-infinity controller presented a satisfactory behavior with smoother responses than the other controllers, however with higher rise times; control devices for the subject controller presented the best transient response among all others and indicated a positive impact in the fuel consumption. Finally, a noise immunity check revealed that this H-infinity controller was capable to properly attenuate high frequency noise normally present in the measurement systems.
