摘要
基于大涵道比涡扇发动机部件级模型,从减少单次流路计算耗时和降低单步流路计算次数两方面研究提高模型实时性的方法。测试并分析了发动机各模块单步计算耗时,通过建立气体热力属性插值表,使模型单次流路计算耗时减少80%,在3.3GHz Intel CPU平台下模型单次流路计算耗时0.02ms,在168MHz STM32F407硬件平台下耗时1.55ms。研究了不同收敛残差对模型流路计算次数及仿真精度的影响。仿真结果表明:相比Newton-Raphson法,Broyden法流路计算次数更少;将迭代求解残差由0.0001调整至0.001或0.005,模型流路计算次数显著减少,低压转速仿真偏差在0.2%以内。
Based on the Component-Level Model(CLM) of a high-bypass-ratio turbofan engine, the ways to improve the real-time performance of CLM were achieved through reducing one flow path calculation time and decreasing the number of one-step calculations. The calculation time of each computation module during one flow path calculation was tested and analyzed. By setting up an interpolation table of gas thermodynamic properties,the one flow path calculation time has been reduced by 80%,which is about 0.02 ms on the platform of3.3GHz Intel CPU and 1.55 ms on that of 168 MHz TM32F407. Besides,the effects of different convergence residuals on the model flow path calculations number and simulation accuracy were compared and analyzed. Simulation results show that the number of model one-step calculations using Broydenmethod is less than that of using Newton-Raphson method.By adjusting the iterative solving residual from 0.0001 to 0.001 or 0.005,the number of model one-step calculations could be significantly reduced while keeping the simulation error of low pressure rotor speed within 0.2%.
出处
《推进技术》
EI
CAS
CSCD
北大核心
2017年第1期199-206,共8页
Journal of Propulsion Technology
基金
国家自然科学基金面上项目(51276087)
中央高校基本科研业务费专项资金(NS2015027)
关键词
涡扇发动机
部件级模型
实时性
流路计算
Broyden法
Turbofan engine
Component-level model
Real-time performance
Flow path calculation
Broyden method