Novel dual-twin model-based nonlinear onboard adaptive modeling method for aircraft engine with fuel measurement uncertainty awareness

The onboard model serves as a digital twin of the real engine, requiring its input parameters to equal those of the engine. However, fuel flow(Wf) is one of the most uncertain parameters to measure, significantly limiting the onboard models' application in engineering. Therefore, this paper proposes a novel dual-twin onboard model architecture that utilizes accurate fan speed(Nf) to perceive measurement uncertainty in Wf. Firstly, a master onboard model with a nonlinear Nf controller is designed. This Nf controller, designed by an online data-driven algorithm, adapts to the engine's nonlinear characteristics. It ensures that the master onboard model can rapidly synchronize with the measured Nf of the engine, enabling real-time online estimation of Wf measurement uncertainty and correction of measured Wf. Considering the master onboard model cannot adapt to engine component performance degradation, a real-time nonlinear onboard adaptive model is established by combining an improved spherical unscented Kalman filter with an auxiliary onboard model, estimating the engine health parameters degradation. These degradation values are processed through outlier-median averaging filtering to update the master onboard model, ensuring accurate correction of measured Wf in the whole lifecycle. To enhance the fault tolerance against Nf sensor faults, a dual-redundancy speed control strategy is proposed to avoid inaccuracies in Wf measurement corrections under single-fan-speed control. Finally, comparative results demonstrate that this novel model accurately estimates Wf measurement uncertainty in all cases, with an improvement of over 60 % compared to existing methods. It exhibits excellent adaptability and fault tolerance, demonstrating significant superiority of the proposed method.