A highly robust thrust estimation method with dissimilar redundancy framework for gas turbine engine. (15th April 2022)

• Record Type: Journal Article
• Authors: Zhao, Hang; Liao, Zengbu; Liu, Jinxin; Li, Ming; Liu, Wei; Wang, Lei; Song, Zhiping
• Is Part Of: Energy. Volume 245(2022)
• Publication Date: 2022-04-15
• Subjects: Gas turbine engine -- Highly robust thrust estimation -- Hybrid framework -- Dissimilar redundancy framework; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2022.123255
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: An accurate thrust estimation method is of great importance for implementing direct thrust control scheme on gas turbine engine in that the in-flight thrust cannot be measured. Meanwhile, some interference factors, such as sensor noise, engine-to-engine variations, engine performance deterioration and changes in atmospheric condition, may have adverse effects on the accuracy of thrust estimation. However, these factors are not fully considered in the design of the existing methods. To ensure the accuracy in practical application, a highly robust thrust estimation (HRTE) method is proposed in this paper. This method is a fusion of three dissimilar hybrid estimation modules, each consisting of a physics-based module and an error-compensated module. Compared with the existing pure data-driven methods, its innovations are as follows: 1) a hybrid framework for thrust estimation is proposed, which helps get rid of the dilemma of learning high-dimensional thrust data and improves the accuracy and robustness. 2) a dissimilar redundancy framework is designed for thrust estimation, which helps improve the adaptivity to those interference factors, especially the tolerance to sensor noise. A series of simulation tests eventually show that the HRTE method has high accuracy and robustness in practical application. Highlights: The proposed thrust estimation method has high accuracy and robustness. The hybrid framework avoids the dilemma of learning high-dimensional thrust data. The dissimilar redundancy framework helps improve the tolerance to sensor noise.