Model-based deduction learning control: A novel method for optimizing gas turbine engine afterburner transient

Abstract

The afterburning phase of an aero gas turbine engine is essential for boosting engine thrust. Traditional methods that combine open-loop afterburner fuel flow with closed-loop nozzle throat area control always degrade control quality during the transients of afterburner activation and deactivation. This results in fluctuations in the turbine outlet total pressure, consequently decreasing the fan surge margin, and may even lead to afterburner ignition failure or fan surge. A model-based deduction learning control method is proposed to address these issues. This method comprises: 1) a model-based offline experience deduction and learning module to enhance the coordination of afterburner fuel flow and nozzle throat area control during the early stages of afterburner activation or deactivation; 2) a power lever angle reference trajectory module designed to enhance the linearity of thrust output; 3) a nonlinear integrated online output module to maintain control stability. Simulation results have shown that the method effectively reduces the fluctuations in turbine outlet total pressure, bolsters the fan surge margin, and improves the linearity of thrust during the afterburning phase.