摘要
针对排气温度裕度(EGTM)数据的非线性、非平稳特征,提出了基于改进经验模态分解(EMD)与支持向量回归结合的预测方法。通过改进的EMD方法对EGTM数据进行分解以降低时间序列的复杂程度;根据EMD得到各本征模态函数及趋势序列,构建基于支持向量机的预测模型;将所得的各分量的预测结果综合得到EGTM预测结果。以某航空发动机EGTM数据验证所提方法的有效性,相比于传统的预测方法,均方根误差降低至2.024,平均绝对位差降低至1.603,有效提高了回归精度。
Aiming at the non-linear and non-stationary features of Exhaust gas temperature margin(EGTM)data,a prediction method based on Improved Ensemble Empirical Mode Decomposition(EMD)and Support Vector Regression(SVR)was proposed.Improved EMD was used to decompose the original EGTM data series to reduce the complexity.According to the Empirical mode decomposition.The Intrinsic mode function and residual series were constructed,and the prediction model based on support vector machine was built.The prediction results were combined to obtain the predicted results of EGTM.The effectiveness of the proposed method was verified by aero engine EGTM data.RMSE and MAE were reduced to 2.024 and 1.603 respectively.Compared with the traditional prediction method,it effectively improves the regression accuracy.
作者
戴邵武
陈强强
丁宇
DAI Shaowu;CHEN Qiangqiang;DING Yu(Naval Aviation University,Yantai 264001,China;Air Force 93381Troop,Lalin 150223,China)
出处
《兵器装备工程学报》
CAS
北大核心
2020年第1期157-162,共6页
Journal of Ordnance Equipment Engineering
基金
山东自然科学基金面上项目(ZR2017MF036)
国防科技项目基金项目(F062102009).
关键词
排气温度裕度
经验模态分解
支持向量机
预测
回归
exhaust gas temperature margin
empirical model decomposition
support vector machine
prediction
regression