航空学报 > 2011, Vol. 32 Issue (8): 1506-1511   doi: CNKI:11-1929/V.20110324.1201.008

基于遗传算法和ARMA模型的航空发电机寿命预测

崔建国1, 赵云龙1, 董世良2, 张红梅1, 陈希成2   

  1. 1. 沈阳航空航天大学 自动化学院, 辽宁 沈阳 110136;
    2. 沈阳飞机设计研究所, 辽宁 沈阳 110035
  • 收稿日期:2010-10-14 修回日期:2010-12-03 出版日期:2011-08-25 发布日期:2011-08-19
  • 通讯作者: 崔建国,Tel.: 024-89724448 E-mail: gordon_cjg@163.com E-mail:gordon_cjg@163.com
  • 作者简介:崔建国(1963-) 男,博士后,教授,研究生导师。主要研究方向:飞行器健康诊断、预测与综合健康管理,仿真技术与应用等。 Tel: 024-89724448 E-mail: gordon_cjg@163.com
  • 基金资助:

    航空科学基金 (2010ZD54012);辽宁省教育厅科研基金 (2008544);国防基础科研计划项目 (A0520110023)

Life Prognostics for Aero-generator Based on Genetic Algorithm and ARMA Model

CUI Jianguo1, ZHAO Yunlong1, DONG Shiliang2, ZHANG Hongmei1, CHEN Xicheng2   

  1. 1. School of Automation, Shenyang Aerospace University, Shenyang 110136, China;
    2. Shenyang Aircraft Design & Research Institute, Shenyang 110035, China
  • Received:2010-10-14 Revised:2010-12-03 Online:2011-08-25 Published:2011-08-19

摘要: 针对航空发电机剩余使用寿命难以准确预测的问题,提出一种基于遗传算法(GA)优化的自回归与移动平均(ARMA)模型。运用航空发电机寿命专用试验平台,对某型航空发电机寿命进行长期试验,获取该航空发电机寿命相关数据。深入分析寿命试验数据,并对其建立ARMA模型。在此基础上,采用遗传算法对ARMA模型的阶数进行优化,分别采用ARMA模型与经遗传算法优化后的ARMA模型对航空发电机的使用寿命进行预测研究。结果表明,优化前后两种ARMA模型均可对航空发电机的使用寿命实现预测效能。优化前ARMA模型对航空发电机寿命预测的平均相对误差为4.33%;而经遗传算法优化后,ARMA模型预测的平均相对误差仅为2.26%,能更准确预测航空发电机的使用寿命,具有很好的工程应用价值。

关键词: 遗传算法, ARMA模型, 航空发电机, 寿命预测, 注油压力

Abstract: To improve the aero-generator life prediction accuracy, an optimized auto-regressive and moving average (ARMA) model based on the genetic algorithm(GA) is presented. A specific experimental platform is used to perform long-term life prediction experiments on a certain type of aero-generator and collect the related test data. After a thorough analysis of these test data, a corresponding ARMA model is designed, and the genetic algorithm is used to carry on the exponent number optimization of the model. Then the original and the optimized ARMA models are used respectively to conduct life prediction research on the service life of the aero-generator. The result shows that these two models can realize the function of predicting the service life of an aero-generator. The average relative prognostic error of the ARMA model after the optimization is 2.26%, which is less than 4.33%, the error of the original model without optimization. Thus a conclusion can be drawn that the optimized ARMA model can predict the service life of an aero-generator more accurately and this model may find wide application in engineering practice.

Key words: genetic algorithm, ARMA model, aero-generator, life prognostics, oil-filled pressure

中图分类号: