航空发动机试车台推力仿真技术研究  

Research on Simulating Test-Bed Thrusts of Aero-Engine

作　　者：宋江涛[1] 邓小宝[1] 魏海涛[1] 

机构地区：[1]中国飞行试验研究院,陕西西安710089

出　　处：《计算机仿真》2012年第1期67-70,共4页Computer Simulation

摘　　要：航空发动机试车准确测量推力问题,航空发动机推力是衡量发动机性能的主要参数。航空发动机试车台推力测量系统复杂,测试环境恶劣,易出现推力测量不准确。为解决上述问题,提出贝叶斯正则化算法的BP神经网络技术和相关性分析建立航空发动机试车台推力模型,进行推力仿真。在发动机试验过程中可实时计算发动机推力,和试车台推力测量系统的测试结果进行对比,判别推力测量系统工作是否正常。仿真结果表明,设计的基于L-M贝叶斯正则化算法的BP神经网仿真模型可准确的计算不同大气条件下和各个工作状态的发动机实际推力。Engine thrust is an important factor evaluating engine performances. Test-bed system is usually com- plex, and test circumstances are bad, therefore, the measured thrust can be inaccuracy. Using a lot of data from actual engine tests as training samples, a model of simulating engine thrust was built based on L-M Bayesian regulation BP neutral networks and relativity analysis. Also this model can be used to real-time simulate engine thrust under different atmospheres and working conditions. Comparing the simulated results with the acutual measured ones, wether the system of measuring thrusts is normal or not can be judged. The simulation results show that the built model in this paper can accurately predict the actual engine thrusts under different temperatures and working conditions.

关 键 词：贝叶斯正则化 线性相关性 试车台 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]