基于粒子群极限学习机的排气温度裕度预测
|
摘要
排气温度是表征发动机工作状态的主要参数之一,通过对多个飞行架次的排气温度裕度(Exhaust Gas Temperature Margin,EGTM)进行预测分析,能够在一定程度上反映发动机工作性能,为后续故障检测工作提供理论依据。针对EGTM数据的非线性、非平稳特征,提出了基于粒子群优化算法(Particle Swarm Optimization,PSO)的极限学习机(Extreme Learning Machine,ELM)预测方法。通过ELM构建EGTM的预测模型,并利用PSO算法对其参数进行优化以保证模型的精确性;以某航空发动机EGTM数据作为验证,结果表明,相比于传统的预测方法,RMSE与MAE分别降低至1.889 8、1.0,有效提高了预测精度。
Exhaust gas temperature is one of the main parameters to reflect the working state of the aero engine. By predicting and analyzing the exhaust gas temperature margin(EGTM) of multiple flight frames, the aero engine performance can be reflected to some extent which provides a theoretical basis for the fault detection work. Aiming at the non-linear and non-stationary features of EGTM data series, an Extreme Learning Machine(ELM) prediction method based on Particle Swarm Optimization(PSO) is proposed. The prediction model of EGTM is constructed by ELM, and its parameters are optimized by PSO algorithm to ensure the accuracy of the model. Verified by the EGTM data of an aero engine, the results show that compared with the traditional prediction method, RMSE and MAE are reduced to 1.8898 and 1.0, which improves the prediction accuracy effectively.
Exhaust gas temperature is one of the main parameters to reflect the working state of the aero engine. By predicting and analyzing the exhaust gas temperature margin(EGTM) of multiple flight frames, the aero engine performance can be reflected to some extent which provides a theoretical basis for the fault detection work. Aiming at the non-linear and non-stationary features of EGTM data series, an Extreme Learning Machine(ELM) prediction method based on Particle Swarm Optimization(PSO) is proposed. The prediction model of EGTM is constructed by ELM, and its parameters are optimized by PSO algorithm to ensure the accuracy of the model. Verified by the EGTM data of an aero engine, the results show that compared with the traditional prediction method, RMSE and MAE are reduced to 1.8898 and 1.0, which improves the prediction accuracy effectively.
引文
[1] 徐亮.航空发动机气路性能参数趋势预测研究[D].广汉:中国民用航空飞行学院,2016.
[2] 曹惠玲,李理,任炎炎,等.航空发动机涡轮结构参数对EGTM的影响研究[J].航空维修与工程,2016(1):29-32.
[3] 李晓白,崔秀伶,郎荣玲.航空发动机性能参数预测方法[J].北京航空航天大学学报,2008,34(3):253-256.
[4] 李书明,马晓政,商智超.基于小波分析的水洗对民航发动机拆发间隔的影响[J].机械工程与自动化,2017(2):144-145.
[5] 罗华柱.基于改进SVM的航空发动机预测方法研究[D].南昌:南昌航空大学,2017.
[6] 刘永建.基于改进神经网络的民机发动机故障诊断与性能预测研究[D].南京:南京航空航天大学,2012.
[7] 林敏.极限学习机在航空发动机气路故障诊断中的应用[D].上海:上海交通大学,2015.
[8] 范伟,田丽,汪晨,等.基于PSO-ELM模型的短期电力系统负荷预测[J].南阳理工学院学报,2017,9(4):12-15.
[9] 王付广,李伟,郑近德,等.基于多频率尺度模糊熵和ELM的滚动轴承剩余寿命预测[J].噪声与振动控制,2018,38(1):188-192.
[10] 王焱,汪震,黄民翔,等.基于OS-ELM和Bootstrap方法的超短期风电功率预测[J].电力系统自动化,2014,38(6):14-19.
[11] 朱兵,许江宁,何泓洋,等.粒子群算法优化的捷联罗经初始对准方法[J].中国惯性技术学报,2017,25(1):47-51.
[12] 任淑红.航空发动机视情维修管理中的寿命预测技术[M].北京:经济科学出版社,2016.
[13] 李冬,李本威,赵凯,等.基于奇异值趋势分解和LS-SVR的航空发动机性能指数组合预测[J].推进技术,2013,34(10):1406-1413.
[14] ZHU B,PICCI G.Proof of local convergence of a new algorithm for covariance matching of periodic ARMA models[J].IEEE Control Systems Letters,2017,1(1):206-211.
[2] 曹惠玲,李理,任炎炎,等.航空发动机涡轮结构参数对EGTM的影响研究[J].航空维修与工程,2016(1):29-32.
[3] 李晓白,崔秀伶,郎荣玲.航空发动机性能参数预测方法[J].北京航空航天大学学报,2008,34(3):253-256.
[4] 李书明,马晓政,商智超.基于小波分析的水洗对民航发动机拆发间隔的影响[J].机械工程与自动化,2017(2):144-145.
[5] 罗华柱.基于改进SVM的航空发动机预测方法研究[D].南昌:南昌航空大学,2017.
[6] 刘永建.基于改进神经网络的民机发动机故障诊断与性能预测研究[D].南京:南京航空航天大学,2012.
[7] 林敏.极限学习机在航空发动机气路故障诊断中的应用[D].上海:上海交通大学,2015.
[8] 范伟,田丽,汪晨,等.基于PSO-ELM模型的短期电力系统负荷预测[J].南阳理工学院学报,2017,9(4):12-15.
[9] 王付广,李伟,郑近德,等.基于多频率尺度模糊熵和ELM的滚动轴承剩余寿命预测[J].噪声与振动控制,2018,38(1):188-192.
[10] 王焱,汪震,黄民翔,等.基于OS-ELM和Bootstrap方法的超短期风电功率预测[J].电力系统自动化,2014,38(6):14-19.
[11] 朱兵,许江宁,何泓洋,等.粒子群算法优化的捷联罗经初始对准方法[J].中国惯性技术学报,2017,25(1):47-51.
[12] 任淑红.航空发动机视情维修管理中的寿命预测技术[M].北京:经济科学出版社,2016.
[13] 李冬,李本威,赵凯,等.基于奇异值趋势分解和LS-SVR的航空发动机性能指数组合预测[J].推进技术,2013,34(10):1406-1413.
[14] ZHU B,PICCI G.Proof of local convergence of a new algorithm for covariance matching of periodic ARMA models[J].IEEE Control Systems Letters,2017,1(1):206-211.