基于多代理模型的航空发动机建模及优化方法

为了进一步提高航空发动机建模及优化方法的性能，本文提出了一种基于多代理模型技术的建模及优化方法。本文首先提出了一种新的代理模型全局误差估计方法，以此建立了新的多代理模型建模方法。然后提出了一种组合模型预测偏差估计方法，以此发展了一种基于多代理模型技术的优化方法。6个不同维度及不同训练集大小的解析测试算例的结果表明，本文所发展的建模方法相较于现有方法精度更高，本文所发展的优化方法相较于经典代理模型优化方法算法收敛性更强。同时变循环发动机稳态性能建模及加速燃油控制规律优化实例表明，本文所发展的方法在处理实际工程问题时，依旧可以表现出良好的算法性能。

Multi-Surrogates Based Modelling and Optimization Algorithm Suitable for Aero-Engine

In order to improve the performance of aero-engine modelling and optimization algorithm, a multi-surrogates based modelling and optimization algorithm is proposed in this paper. First, a new surrogate global error measure named Mean-Variance Error (MVE) is proposed. Then an average ensemble model is built based on the MVE. An ensemble prediction uncertainty estimator named Ensemble Prediction Variance (EPV) is also proposed in this paper. Then a sequential ensemble optimization algorithm is developed based on the EPV. By using six well-known mathematical functions with varying dimensions and numbers of training points, it is proved that the proposed ensemble model is more accurate than the other ensemble models, and the convergence of the proposed optimization algorithm is better than that of the classic optimization algorithm. Meanwhile, the steady performance modelling problem of the variable cycle engine and the optimization problem of the variable cycle engine acceleration fuel control schedule are also considered, it is proved that the proposed algorithms perform well in solving a complex engineering problem.