基于NARMAX辨识航空发动机起动过程数学模型  

Identification of Aero-engine Starting Process Mathematical Model Based on NARMAX

作　　者：郭政波 望佳 尹文强[1] 姚尚宏[1] 苏三买[2] 

机构地区：[1]中国飞行试验研究院发动机所,陕西西安710089 [2]西北工业大学动力与能源学院,陕西西安710072

出　　处：《航空科学技术》2016年第5期22-26,共5页Aeronautical Science & Technology

摘　　要：建立准确的数学模型对于航空发动机的设计开发有十分重要的作用。航空发动机是一种非常复杂的非线性系统,基于外源输入下的非线性自回归滑动平均模型(NARMAX)可以很好地逼近任意非线性。为此,提出了采用改进的直交化最小二乘法并结合误差下降速率(ERR)算法,辨识航空发动机起动过程的NARMAX模型。结果表明,与最小二乘法辨识得到的结果相比,航空发动机起动过程的NARMAX模型更准确,收敛速度更快,能够满足发动机模型的精度与实时性要求,可以实时反映发动机性能。Developing a mathematical model with high accuracy is of great importance to the design and development of aero-engine. The aero-engine is a complex nonlinear system. Nonlinear Auto Regressive Moving Average model with exogenous input （NARMAX） is able to model any kinds of nonlinearities. As a result, the aero-engine starting process was simulated by a NARMAX model utilizing an improved orthogonal least squares with error reduction ratio algorithm. According to the experimental results, the developed NARMAX model has a better accuracy and faster regression than the least squares method for identifying the aero-engine starting process. Therefore, the NARMAX model can meet the accuracy and real-time requirements, and thus capture the engine performance effectively in real-time.

关 键 词：航空发动机 辨识模型 NARMAX模型 ERR算法 起动过程 

分 类 号：V233.2[航空宇航科学与技术—航空宇航推进理论与工程]