北京航空航天大学学报 ›› 2018, Vol. 44 ›› Issue (2): 305-315.doi: 10.13700/j.bh.1001-5965.2017.0078

• 论文 • 上一篇    下一篇

基于QPSO混合算法的变循环发动机建模方法

肖红亮1, 李华聪1, 李嘉1, 王淑红2, 彭凯1   

  1. 1. 西北工业大学 动力与能源学院, 西安 710072;
    2. 中国航发西安动力控制科技有限公司, 西安 700077
  • 收稿日期:2017-02-21 修回日期:2017-06-05 出版日期:2018-02-20 发布日期:2018-03-01
  • 通讯作者: 李华聪,E-mail:lihuacong@nwpu.edu.cn E-mail:lihuacong@nwpu.edu.cn
  • 作者简介:肖红亮,男,博士研究生。主要研究方向:航空发动机建模及控制;李华聪,男,博士,教授,博士生导师。主要研究方向:航空发动机控制、先进控制理论。
  • 基金资助:
    国家自然科学基金(51506176);航空科学基金(6141B090302);中央高校基本科研业务费专项资金(G2017KY0003)

Modeling method of variable cycle engine based on QPSO hybrid algorithm

XIAO Hongliang1, LI Huacong1, LI Jia1, WANG Shuhong2, PENG Kai1   

  1. 1. School of Power and Energy, Northwestern Polytechnical University, Xi'an 710072;
    2. AVIC Xi'an Aero-engine Controls Technology Co., Ltd., Xi'an 700077, China
  • Received:2017-02-21 Revised:2017-06-05 Online:2018-02-20 Published:2018-03-01
  • Supported by:
    National Natural Science Foundation of China (51506176);Aeronautical Science Foundation of China (6141B090302);the Fundamental Research Funds for the Central Universities (G2017 KY0003)

摘要: 针对变循环发动机非线性部件模型共同工作方程组求解时初值选取对收敛速度和精度的影响问题,提出一种基于量子粒子群优化(QPSO)算法与Broyden拟牛顿法混合的求解思路。首先,对变循环发动机(VCE)进行变几何特性分析以及反向传播(BP)神经网络下的外涵道稳态特性分析基础上,建立反映变几何特性以及模式切换等全状态部件模型。其次,以该模型性能计算为基准,提出了一种基于QPSO的Broyden拟牛顿混合算法来达到发动机共同工作平衡要求,通过发散系数实现混合算法的切换,以改善单一Broyden拟牛顿法对初值选取的依赖性同时提高QPSO算法的求解效率。通过高阶非线性方程组的仿真验证了算法的有效性、求解效率以及精度。最后,进行VCE部件模型稳态、动态仿真计算,结果表明:与GasTurb性能计算结果对比可以看出发动机速度特性、高度特性等变化趋势与GasTurb基本一致,且误差均小于2%;基于QPSO的Broyden拟牛顿混合算法可有效快速地完成VCE部件模型的求解;所建VCE部件模型能够有效实现该新型发动机的性能模拟分析。

关键词: 变循环发动机(VCE), 变几何特性, 外涵道, 非线性方程组求解, 量子粒子群优化(QPSO), Broyden拟牛顿法

Abstract: A new hybrid algorithm which is based on quantum particle swarm optimization (QPSO) algorithm and Broyden quasi-Newton algorithm was proposed to reduce the effect of initial value selection on convergence speed and accuracy in solving the variable cycle engine (VCE) model. Firstly, based on the analysis of the VCE geometrical characteristics and the analysis of the steady-state characteristics of the external duct through backpropagation(BP) neural network method, a component model was established which can reflect variable geometry property and mode switching and other states of the VCE. Secondly, based on the model performance calculation, a QPSO based Broyden quasi-Newton hybrid algorithm was used to solve the VCE model cooperating equations, which improved the convergence and calculation efficiency of the hybrid algorithm by introducing the divergence coefficient to combine the two single algorithms. The effectiveness, efficiency and accuracy of the algorithm were verified by the simulation of high-order nonlinear equations. Finally, the steady state and dynamic simulation of VCE component model were carried out. The results of VCE model show that, compared with the results of GasTurb performance calculation, the trends of velocity characteristics and altitude characteristics are basically the same with those of GasTurb, the error between VCE model and GasTurb is less than 2%. The hybrid algorithm based on QPSO and Broyden quasi-Newton algorithm can solve the VCE model efficiently and quickly. The established VCE model can be used for performance simulation and analysis.

Key words: variable cycle engine (VCE), variable geometry characteristics, external duct, solving nonlinear equations, quantum particle swarm optimization (QPSO), Broyden quasi-Newton method

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发