基于微分进化算法的航空发动机模型修正  

Correction of Aeroengine Model Based on Differential Evolution Algorithm

作　　者：朱正琛 李秋红[1] 王元[1] 潘鹏飞[1] 

机构地区：[1]南京航空航天大学江苏省航空动力系统重点实验室,南京210016

出　　处：《航空发动机》2016年第1期53-58,共6页Aeroengine

基　　金：国家重大基础研究项目资助

摘　　要：为了提高航空发动机性能仿真模型精度,采用微分进化算法对发动机部件特性进行修正。对微分进化算法进行改进,提出折线式交叉变量变化方式,提高了算法的寻优能力。提出变步长牛顿-拉夫逊迭代算法,基于平衡方程残差范数变化趋势,改变牛顿-拉夫逊算法迭代计算步长,提高了模型的收敛性和收敛速度。在设计点,对各部件特性、引气系数、总压恢复系数进行修正,使修正后的模型输出与试验数据相匹配。仿真结果表明:改进后的牛顿-拉夫逊迭代算法收敛性更强、计算速度更快,修正后的各输出参数的最大建模误差减小到1.3762%,满足建模误差需求。In order to enhance the accuracy of the model, Differential Evolution（DE） algorithm was used to modify the component characteristics of an aeroengine. DE algorithm was improved, the broken line style across-variable was put forward, the optimization ability of the algorithm was increased. A variable step-size Newton-Raphson iteration algorithm was proposed based on the variation tendency of the residue errors norm of the balance equations, which could adjust step-size of Newton-Raphson resulting in improving the convergent ability and convergent speed of the model. At the design point, the characteristics correction coefficients of components, air-entraining correction coefficients and pressure recovery correction coefficients were optimized to achieve high matching accuracy of engine model outputs to test data. The simulation results show that the model based on variable step-size Newton-Raphson method could achieve better convergent performance with less time. After correction of aeroengine model, the maximum error was reduced to 1.3762 %,which satisfied the modeling requirement.

关 键 词：微分进化算法 牛顿-拉夫逊迭代算法 部件特性修正 折线式交叉变量 性能仿真模型 航空发动机 

分 类 号：V233.7[航空宇航科学与技术—航空宇航推进理论与工程]