基于NARX神经网络航空发动机参数动态辨识模型

发布时间：2018-02-04 20:18

  本文关键词： 航空发动机 动态模型 非线性系统辨识 NARX网络　出处：《计算机工程与应用》2017年12期 　论文类型：期刊论文

【摘要】：针对航空发动机参数非线性动态特性,提出一种基于外部输入非线性自回归(NARX)神经网络的发动机参数动态辨识模型。主要思路是根据NARX网络的非线性时序预测特性,结合发动机参数的稳态和动态参数,提出一种基于偏稳态差值预测的NARX参数动态模型结构。设计了SP-P辨识结构,整定了模型内部结构参数并建立N1(低压转子转速)、N2(高压转子转速)、EGT(涡轮后排气温度)参数非线性差分预测模型。最后依据某发动机试车样本,对推杆加减速时N1、N2、EGT动态辨模型进行仿真。仿真结果表明,N2相对误差小于0.2%,N1相对误差小于0.3%,EGT相对误差小于1℃,满足发动机试车仿真需要。最后,将所建模型应用于某A320机务维修训练器的发动机仿真系统。
[Abstract]:Aiming at the nonlinear dynamic characteristics of aero-engine parameters. A dynamic identification model of engine parameters based on external input nonlinear autoregressive neural network is proposed. The main idea is based on the nonlinear prediction characteristics of NARX neural network. Combining the steady and dynamic parameters of engine parameters, a dynamic model structure of NARX parameters based on partial steady-state difference prediction is proposed, and the SP-P identification structure is designed. The internal structural parameters of the model were set up and the N1 (low pressure rotor speed) was established. EGT (turbine exhaust temperature) parameter nonlinear differential prediction model. Finally, according to a test sample of an engine, the dynamic identification model of N _ (1) N _ (2) EGT during acceleration and deceleration of push rod is simulated. The simulation results show that. The relative error of N2 is less than 0.2 and N1 is less than 0.3 and EGT relative error is less than 1 鈩，

本文编号：1491061