Koopman-Based Nonlinear Identification and Model Predictive Control of a Turbofan Engine

David Grasev

cs.LGApr 2, 2026

Koopman-Based Nonlinear Identification and Model Predictive Control of a Turbofan Engine

Authors: David Grasev

Organizations: Department of Aviation Technology, University of Defence, Brno, Czech Republic.

Abstract

This paper investigates Koopman operator-based approaches for multivariable control of a two-spool turbofan engine. A physics-based component-level model is developed to generate training data and validate the controllers. A meta-heuristic extended dynamic mode decomposition is adapted, with a cost function designed to accurately capture both spool-speed dynamics and the engine pressure ratio (EPR), enabling the construction of a single Koopman model that can be reused across multiple control strategies. Using the identified time-varying Koopman model, an adaptive Koopman-based model predictive controller (AKMPC) with a disturbance observer is developed and compared with a Koopman-based feedback linearization controller (K-FBLC) and its integrator-augmented version (K-FBLC-I). The Koopman representation further enables nonlinear GTE output limiters, such as rotor-acceleration and turbine-inlet-temperature limits, to be expressed as linear constraints in the AKMPC. The controllers are evaluated for two control configurations of spool speeds and EPR, under both sea-level and varying flight conditions. The results demonstrate that the proposed identification approach enables accurate predictions of both spool speeds and EPR, allowing the Koopman model to be reused flexibly across different control formulations. While all strategies achieve comparable performance in sea-level conditions, the AKMPC demonstrates improved performance under varying flight conditions due to its ability to capture nonlinear dynamics, handle constraints, and compensate for model mismatch. Moreover, the EPR control strategy improves the thrust response. The study highlights the applicability of Koopman-based control and the advantages of the AKMPC framework for robust turbofan engine control.

Figures & tables

Symbol	Dimension / unit	Description
$N_{1},N_{2}$	RPM	Low- and high-pressure spool speeds
$W_{\mathrm{f}}$	kg/s	Fuel flow
$A_{\mathrm{n}}$	m 2	Nozzle area
$\Pi_{\mathrm{EPR}}$	–	Engine pressure ratio
$F$	N	Net thrust
$\mathbf{x}$	$\mathbb{R}^{2}$	State vector, $\mathbf{x}=[N_{1},N_{2}]^{\top}$

Table 1: Summary of the main notation used in the manuscript.

Figure 1: Turbofan GTE diagram with the main components considered in this paper.

Figure 2: Schematic diagram of Koopman eigenfunction lifting and evolution.

Figure 3: The AKMPC diagram showing the interaction between the Koopman system and MPC optimization.

Panel of Figure 4 — Figure 4: (a) Thrust and (b) fuel flow steady-state validation against the GasTurb 15 simulation software in $H=0$ km, $M_{0}=0$ and $H=10$ km, $M_{0}=0.8$ .

Steady-state	$H=0$ km, $M_{0}=0$	$H=10$ km, $M_{0}=0.8$
$\mathrm{MAE_{F}}$	1010.4 N	403.75 N
$\mathrm{MAPE_{F}}$	4.49 $\%$	2.37 $\%$
$\mathrm{MAE_{WF}}$	0.0183 kg/s	0.0031 kg/s
$\mathrm{MAPE_{WF}}$	2.86 $\%$	1.42 $\%$

Table 2: MAE and MAPE values for the steady-state validation.

Figure 5: Transient performance validation against the GasTurb 15 simulation software for the sea-level conditions. The fuel flow is rescaled for better visualization.

Transient	$H=0$ km, $M_{0}=0$
$\mathrm{MAE_{N1}}$	83.7 RPM
$\mathrm{MAPE_{N1}}$	0.744 $\%$
$\mathrm{MAE_{N2}}$	51.2 RPM
$\mathrm{MAPE_{N2}}$	0.274 $\%$

Table 3: MAE and MAPE values for the transient validation.

Figure 6: Dataset used for identification of the 2-spool GTE (without noise for clarity).

Function	LF	GAU	IQ
$\mathrm{MAE_{N1}}$ (RPM)	31.92	19.14	20.3
$\mathrm{MAE_{N2}}$ (RPM)	30.58	17.05	15.17
$\mathrm{MAE_{EPR}}$	0.0182	0.0116	0.0114
$\mathrm{MAPE_{N1}}$ $\%$	1.52	0.656	0.68
$\mathrm{MAPE_{N2}}$ $\%$	1.76	0.55	0.6
$\mathrm{MAPE_{EPR}}$ $\%$	2.2	1.21	1.12

Table 4: Comparison of the observable functions for MH-EDMD.

Figure 7: Prediction results of the identified Koopman system after numerical integration.

Metric	EDMDc, GAU	MH-EDMD, IQ
$\mathrm{MAE}_{N_{1}}$ (RPM)	372.3	20.3
$\mathrm{MAE}_{N_{2}}$ (RPM)	433.7	15.17
$\mathrm{MAE}_{\mathrm{EPR}}$	0.0523	0.0114

Table 5: Comparison of the MH-EDMD with the EDMDc with Gaussian RBF.

Figure 8: Spool speed response in sea-level conditions.

Panel of Figure 9 — Figure 9: (a) Fuel flow and (b) nozzle area comparison for $N_{1}-N_{2}$ in sea-level conditions.

Controller	$\mathrm{IAE}\ N_{1}$	$\mathrm{IAE}\ N_{2}$
Unit	$\mathrm{RPM\,s}$	$\mathrm{RPM\,s}$
PI	17381	14600
K-FBLC	16763	13526
K-FBLC-I	15740	12660
AKMPC	14869	11744

Table 6: The IAE criterion values for the controllers in sea-level conditions.

Figure 10: Estimated disturbances for the

N_{1}-N_{2}

case in sea-level conditions.

Panel of Figure 11 — Figure 11: (a) Altitude and (b) Mach number of flight for the evaluation.

Figure 12: Spool speed response in varying flight conditions.

Figure 13: Comparison of fuel flow with limits in varying flight conditions.

Controller	$\mathrm{IAE}\ N_{1}$	$\mathrm{IAE}\ N_{2}$
Unit	$\mathrm{RPM\,s}$	$\mathrm{RPM\,s}$
PI	24632	22621
K-FBLC	27971	21807
K-FBLC-I	25760	20710
AKMPC	23835	19548

Table 7: The IAE criterion values for the controllers in varying flight conditions.

Panel of Figure 14 — Figure 14: (a) Spool speeds, (b) fuel flow, and (c) nozzle area comparison for AKMPC with and without the DO in varying flight conditions.

Figure 15: Estimated disturbances for the

N_{1}-N_{2}

case in varying flight conditions.

Item	Specification
CPU	AMD Ryzen 7 8845HS 3.8 GHz
RAM	32 GB DDR5
Operating System	Microsoft Windows 11 10.0.26200
MATLAB Version	R2025b Update 4

Table 8: Hardware and software specifications.

Case	Mean (ms)	STD (ms)	Median (ms)
SL	0.94	0.27	0.86
VFCs	0.98	0.30	0.88

Table 9: The per-step computational times for the AKMPC and Koopman predictor.

Panel of Figure 16 — Figure 16: Effect of the TIT and $\dot{N}_{1}$ limiters on the limited outputs - (a) the $\dot{N}_{1}$ and (b) the TIT. The limiters successfully prevented the system from exceeding the set operational limits.

Figure 17: The relation between the corrected thrust and engine pressure ratio across varying Mach numbers.

Figure 18: EPR and

N_{1}

response in sea-level conditions. Black circles mark K-FBLC and K-FBLC-I overshoots.

Panel of Figure 19 — Figure 19: (a) Fuel flow and (b) nozzle area comparison for $\Pi_{\mathrm{EPR}}-N_{1}$ in sea-level conditions.

Controller	$\mathrm{IAE}\ N_{1}$	$\mathrm{IAE}\ \Pi_{\mathrm{EPR}}$
Unit	$\mathrm{RPM\,s}$	(-)
K-FBLC	21380	3.88
K-FBLC-I	21400	3.785
AKMPC	20922	3.738

Table 10: The IAE criterion values for the controllers in sea-level conditions.

Figure 20: Estimated disturbances for the

\Pi_{\mathrm{EPR}}-N_{1}

case in sea-level conditions.

Figure 21: Thrust comparison for the two control strategies. Black circles mark regions with excessive oscillations for the

N_{1}-N_{2}

controller.

Figure 22: EPR and

N_{1}

response in varying flight conditions. Black circles mark the regions with offset and overshoots.

Panel of Figure 23 — Figure 23: (a) Fuel flow and (b) nozzle area comparison for $\Pi_{\mathrm{EPR}}-N_{1}$ in varying flight conditions.

Controller	$\mathrm{IAE}\ N_{1}$	$\mathrm{IAE}\ \Pi_{\mathrm{EPR}}$
Unit	$\mathrm{RPM\,s}$	(-)
K-FBLC	25366	4.703
K-FBLC-I	24500	4.55
AKMPC	23807	4.331

Table 11: The IAE criterion values for the controllers in varying flight conditions.

Figure 24: Estimated disturbances for the

\Pi_{\mathrm{EPR}}-N_{1}

case in varying flight conditions.

Extended Dynamic Mode Decomposition

See all 8 papers

Jul 2, 2026

Koopman operator theory: fundamentals, control, and applications

The Koopman operator has gained considerable attention due to its ability to provide a global linear representation of highly complex dynamical systems. The operator describes nonlinear dynamics in a linear way through the lens of real- or complex-valued observable functions. Data-driven techniques, like extended dynamic mode decomposition (EDMD), kernel EDMD, and machine-learning methods, can be used to generate finite-dimensional approximations accompanied by finite-data error bounds. In this tutorial paper, we provide a concise introduction into Koopman operator theory and its use in systems and control. A particular focus is put on data-driven surrogate models, their extension to systems with inputs, and controller design using Koopman operator theory. Moreover, we demonstrate the key techniques, i.e., EDMD and Koopman MPC. To this end, we provide simulation studies including source code on GitHub to enable the interested reader to experience the Koopman operator in systems and control step by step.

Igor Mezić, Jorge Cortés, Karl Worthmann +2

Department of Mechanical Engineering, University of California, Santa Barbara, USA · Department of Mechanical and Aerospace Engineering, University of California, San Diego, USA · Optimization-based Control Group, TU Ilmenau, Germany +2

Surrogate Modeling Koopman Operator Learning Latent Dynamics Modeling Extended Dynamic Mode Decomposition Data-Driven Control

Ask PDF

Mar 23, 2026

Conformal Koopman for Embedded Nonlinear Control with Statistical Robustness: Theory and Real-World Validation

We propose a fully data-driven, Koopman-based framework for statistically robust control of discrete-time nonlinear systems with linear embeddings. Establishing a connection between the Koopman operator and contraction theory, it offers distribution-free probabilistic bounds on the state tracking error under Koopman modeling uncertainty. Conformal prediction is employed here to rigorously derive a bound on the state-dependent modeling uncertainty throughout the trajectory, ensuring safety and robustness without assuming a specific error prediction structure or distribution. Unlike prior approaches that merely combine conformal prediction with Koopman-based control in an open-loop setting, our method establishes a closed-loop control architecture with formal guarantees that explicitly account for both forward and inverse modeling errors. Also, by expressing the tracking error bound in terms of the control parameters and the modeling errors, our framework offers a quantitative means to formally enhance the performance of arbitrary Koopman-based control. We validate our method both in numerical simulations with the Dubins car and in real-world experiments with a highly nonlinear flapping-wing drone. The results demonstrate that our method indeed provides formal safety guarantees while maintaining accurate tracking performance under Koopman modeling uncertainty.

Koki Hirano, Hiroyasu Tsukamoto

Department of Aerospace Engineering, The Grainger College of Engineering, University of Illinois Urbana-Champaign, Urbana, Illinois 61801

Robust Control Feedback Control Robot Trajectory Tracking

Ask PDF

May 12, 2024

Nonparametric Control Koopman Operators

This paper presents a novel Koopman composition operator representation framework for control systems in reproducing kernel Hilbert spaces (RKHSs) that is free of explicit dictionary or input parametrizations. By establishing fundamental equivalences between different model representations, we are able to close the gap of control system operator learning and infinite-dimensional regression, enabling various empirical estimators and the connection to the well-understood learning theory in RKHSs under one unified framework. Consequently, our proposed framework allows for arbitrarily accurate finite-rank approximations in infinite-dimensional spaces and leads to finite-dimensional predictors without a priori restrictions to a finite span of functions or inputs. To enable applications to high-dimensional control systems, we improve the scalability of our proposed control Koopman operator estimates by utilizing sketching techniques. Numerical experiments demonstrate superior prediction accuracy compared to bilinear EDMD, especially in high dimensions. Finally, we show that our learned models are readily interfaced with linear-parameter-varying techniques for model predictive control.

Petar Bevanda, Bas Driessen, Lucian Cristian Iacob +3

Chair of Information-oriented Control, TU München, Germany · Control Systems Group, TU/e, Eindhoven, The Netherlands · Systems and Control Laboratory, HUNREN Institute for Computer Science and Control, Budapest, Hungary +2

Koopman Operator Learning System Identification Kernel Methods

Ask PDF

Abstract

Extended Dynamic Mode Decomposition

Explore similar work

Koopman operator theory: fundamentals, control, and applications

Conformal Koopman for Embedded Nonlinear Control with Statistical Robustness: Theory and Real-World Validation

Nonparametric Control Koopman Operators