Multivariable Adaptive LQ Control of Jet Engines

This paper develops an observer-based indirect adaptive linear quadratic ( LQ) control scheme for a class of MIMO LTI models of jet engine instabilities. These models contain uncertain time delays and resonant modes, leading to a non-minimum phase system characterization which prevents the use of more common direct adaptive schemes such as model reference adaptive control, and adaptive backstepping. The LQ control scheme has the desired optimality for jet engine control applications, in particular, to avoid large transients in the control input which may lead to problems such as actuator saturation. By imposing a high cost of control in the cost function, an LQ control law can keep the control input within the desired operating range while maintaining stability and achieving system performance. To deal with system parameter uncertainties, adaptive parameter estimates are used to solve an online Riccati equation to calculate the control gain matrix. Simulation results have verified the desired performance.