本文選題：近紅外可見光譜 + 十六烷值　； 參考：《光譜學與光譜分析》2017年06期

【摘要】：快速測量十六烷值對檢測柴油品質及控制煉制工藝具有重大意義。首先對采集到的381份柴油樣品進行近紅外可見光譜波段全光譜掃描,利用小波分析(WT)對原始數據進行去噪聲處理,應用競爭性自適應重加權算法(CARS)進行特征波長選擇,將CARS提取的22個特征波長輸入至LS-SVM預測模型,決定系數r2為0.723,預測均方根誤差RMSEP為1.878%。結果表明,使用WT-CARS變量選擇算法獲取光譜特征波長,結合LS-SVM建模,可以快速、準確的測量柴油中的十六烷值,為進一步實現柴油十六烷值的在線檢測以及其他性能參數的快速測定奠定了基礎。
[Abstract]:Rapid measurement of cetane number is of great significance for diesel oil quality detection and refining process control. First, 381 samples of diesel oil were scanned in the near infrared visible spectrum band. The original data were processed by wavelet analysis (WTT), and the characteristic wavelength was selected by competitive adaptive reweighting algorithm (CARSs). The 22 characteristic wavelengths extracted by CARS were inputted into the LS-SVM prediction model, the determination coefficient R2 was 0.723, and the root mean square error (RMSEP) of prediction was 1.878. The results show that using WT-CARS variable selection algorithm to obtain spectral characteristic wavelength and LS-SVM modeling can quickly and accurately measure the cetane number in diesel oil. It lays a foundation for further on-line detection of cetane number of diesel oil and rapid determination of other performance parameters.
【作者單位】： 東華大學機械工程學院;臺州學院機械工程學院;華東交通大學電氣工程與自動化學院;
【基金】：國家自然科學基金項目(61565005) 江西省科技支撐項目(20161BAB202060,20161BBF60060)資助
【分類號】：O657.33;TE626.24

【相似文獻】

相關會議論文 前1條

1 張金生;李麗華;;PLS-NIR分光光度法預測柴油十六烷值[A];全國第10屆分子光譜學術報告會論文集[C];1998年

，

本文編號：1932709