A comparative study of LPV modeling for turbofan engines

Over the past decades, the linear parameter-varying (LPV) strategy has gained a wide range of attention from both academia and industry. Unfortunately, most of these works originate in the context of control engineering applications. Currently, there is no unanimous agreement on which scheduling parameters play a much more important role in the accuracy of LPV-based turbofan engine modeling. The motivation of this paper is to make an evaluation of the impact of different scheduling parameters upon modeling turbofan engines. In this paper, a quasi-LPV model of some commercial turbofan engines is derived from the use of the linearized Jacobian method and the selection of scheduling parameters is discussed. First of all, this paper makes a comparison between three different scheduling parameters|N1, N2, Euclidean norm of N1 and N2 for LPV modeling at the ground condition. Numerical simulation illustrates the advantage of the Euclidean norm of N1 and N2. To accompolish the full-envelope LPV modeling of the concerned turbofan engine, this paper further takes the advantage of the Euclidean norm of height, Mach number and high-pressure rotor speed as the scheduling parameter. The effectiveness of the LPV model at ground and air conditions is validated through the benchmark of a high-fidelity nonlinear component-level engine model. Experimental results reveal that the proposed model gives a perfect accuracy in steady-state conditions and that the relative errors in transient conditions are within 3%.