舰船电子工程
艦船電子工程
함선전자공정
Ship Electronic Engineering
2015年
10期
138-141
,共4页
50℃ 运动粘度%近红外光谱%偏最小二乘法%BP人工神经网络
50℃ 運動粘度%近紅外光譜%偏最小二乘法%BP人工神經網絡
50℃ 운동점도%근홍외광보%편최소이승법%BP인공신경망락
50 ℃ viscosity%near infrared spectroscopy%partial least squares%Back-Propagation network
运动粘度是航空液压油重要的理化性能指标之一，传统理化检测方法费时费力。论文主要通过近红外光谱分析技术实现了运动粘度指标的简便、快速、无损检测。文中比较了平滑、导数、均值中心化和正交信号处理四种不同预处理方法的单独和结合处理的优劣，建立了偏最小二乘法（PLS）模型和 PLS 与误差反向传播神经网络相结合（PLS‐BP）的模型，将预测相关系数（RC）和标准校正偏差（SEC）作为模型主要评价指标，两种模型均通过验证，RC 分别为0．978和0．990。比较结果表明，PLS — BP 模型明显优于 PLS 模型，预测精度较高。
運動粘度是航空液壓油重要的理化性能指標之一，傳統理化檢測方法費時費力。論文主要通過近紅外光譜分析技術實現瞭運動粘度指標的簡便、快速、無損檢測。文中比較瞭平滑、導數、均值中心化和正交信號處理四種不同預處理方法的單獨和結閤處理的優劣，建立瞭偏最小二乘法（PLS）模型和 PLS 與誤差反嚮傳播神經網絡相結閤（PLS‐BP）的模型，將預測相關繫數（RC）和標準校正偏差（SEC）作為模型主要評價指標，兩種模型均通過驗證，RC 分彆為0．978和0．990。比較結果錶明，PLS — BP 模型明顯優于 PLS 模型，預測精度較高。
운동점도시항공액압유중요적이화성능지표지일，전통이화검측방법비시비력。논문주요통과근홍외광보분석기술실현료운동점도지표적간편、쾌속、무손검측。문중비교료평활、도수、균치중심화화정교신호처리사충불동예처리방법적단독화결합처리적우렬，건립료편최소이승법（PLS）모형화 PLS 여오차반향전파신경망락상결합（PLS‐BP）적모형，장예측상관계수（RC）화표준교정편차（SEC）작위모형주요평개지표，량충모형균통과험증，RC 분별위0．978화0．990。비교결과표명，PLS — BP 모형명현우우 PLS 모형，예측정도교고。
Kinematic viscosity is one of the most important Physico‐chemical performance indexes of aviation hydraulic oil ,and the traditional physical and chemical detection method is time consuming .This article mainly involves how to achieve the kinematic viscosity index's simple ,rapid and nondestructive testing through near‐infrared spectroscopy technology .It compares four different pretreatment methods of single and combined with the merits of the processing , including the smoothed ,derivative ,mean centered and orthogonal signal correction .And the Partial Least Squares (PLS) and PLS models with error back propagation neural network combining (PLS‐BP) model are established .The prediction coefficient (RC) and the standard deviation of calibration (SEC) are regarded as their main evaluation indexes .These two models are all verified , RC is 0 .978 and 0 .990 respectively .The result of comparison shows that ,PLS‐BP model is superior to PLS model ,and the accuracy of prediction is higher .
