To enhance the economic efficiency and reliability of condition-based maintenance for aero-engines, it is crucial to achieve accurate quantitative assessment of component health. Addressing the current limitations of existing methods—such as constrained assessment levels, insufficient accuracy, and weak physical interpretability—this paper proposes a component degradation assessment framework based on the Engine Physics-Guided Optimization (EPGO) algorithm. Aimed at high-precision extraction of degradation factors, the framework transforms the complex relationship between component degradation factors and multi-sensor responses into a computable, physics-guided signal by constructing a linear influence matrix. It dynamically generates an ‘engine-guide’ vector to achieve precise and stable extraction of degradation factors within the complex parameter space. To validate the effectiveness of the method, a turboshaft engine is selected as the research object, and both simulation data and actual flight data are employed to verify the core mechanism and engineering applicability of the algorithm. Verification results using simulation data show that, compared with the traditional particle swarm optimization algorithm, the EPGO-based method reduces the average root mean square error of extracted component degradation factors by 36.69% and increases the average correlation coefficient by 29.75%. Results from flight data indicate that the degradation trends revealed by the factors extracted via the EPGO algorithm align with the physical mechanisms of component performance degradation, demonstrating the algorithm's good engineering interpretability and robustness in degradation factor extraction. This study provides a high-precision and highly reliable solution for aero-engine component degradation assessment.

Quick and accurate determination of the optimal synchrophase angle is crucial for synchrophasing control of multi-propeller aircraft with low noise. This paper proposes a novel noise prediction and optimization strategy, developing a continuous and accurate noise prediction model and obtaining its minimum by solving the Hessian matrix and Fourier-Frobenius matrix. Firstly, a novel propeller noise prediction method uses acoustic simulation pressure signals and improved propeller signatures theory to accurately estimate noise for all synchrophase angles and receiving points. Secondly, a novel optimization approach is proposed to solve the analytical solution of the minimum propeller noise: (A) A noise objective function is established, and use its first derivatives’ zeros and Hessian matrix to determine the function minimum. (B) A novel Euler formula transform method is proposed to convert trigonometric polynomials into algebraic polynomials, changing the zeros of the former into those of the latter. (C) Utilize the Fourier-Frobenius matrix method to solve the zeros of algebraic polynomials. To assess the computation time and accuracy, a turboprop aircraft with two six-bladed propellers was analyzed using the computational fluid dynamics and acoustic analogy method, providing acoustic pressure signals at 20 receivers for noise prediction and optimization. The Durand-Kerner and Fourier-Frobenius matrix methods were compared. Results demonstrate that improved propeller signatures theory is more accurate, and the Hessian matrix + Fourier-Frobenius matrix method is faster and more precise than the Hessian matrix + Durand-Kerner method.

In this paper, variable-weights neural network is proposed to construct variable cycle engine’s analytical redundancy, when all control variables and environmental variables are changing simultaneously, also accompanied with the whole engine’s degradation. In another word, variable-weights neural network is proposed to solve a multi-variable, strongly nonlinear, dynamic and time-varying problem. By making weights a function of input, variable-weights neural network’s nonlinear expressive capability is increased dramatically at the same time of decreasing the number of parameters. Results demonstrate that although variable-weights neural network and other algorithms excel in different analytical redundancy tasks, due to the fact that variable-weights neural network’s calculation time is less than one fifth of other algorithms, the calculation efficiency of variable-weights neural network is five times more than other algorithms. Variable-weights neural network not only provides critical variable-weights thought that could be applied in almost all machine learning methods, but also blazes a new way to apply deep learning methods to aeroengines.