獨立成分分析法之蓮霧糖度近紅外光定量分析   全文下載

Integration of Independent Component Analysis with Near Infrared Spectroscopy for Rapid Quantification of Sugar Content in Wax Jambu (Syzygium samarangense Merrill & Perry)

本研究結合獨立成分分析法與近紅外光光譜於蓮霧（Syzygium samaran-gense Merrill & Perry）糖度之快速定量分析，結合JADE演算法、線性迴歸及光譜前處理方法，分別對蓮霧與蔗糖溶液之近紅外光光譜進行分析。相較於其他多變量分析方法，獨立成分分析法可提供更完整之蓮霧糖度資訊，其最佳光譜檢量模式使用一次微分光譜搭配正規化處理，光譜範圍為600至700 nm與900至1098 nm，獨立成分數目為9，rc = 0.988，SEC = 0.243°Brix，rv = 0.971，SEV = 0.381°Brix，RPD = 4.15。獨立成分分析法於蓮霧與蔗糖溶液之分析結果皆較PLSR為佳，顯示獨立成分分析法可快速準確地擷取蓮霧光譜中之糖度資訊，並建立具高預測能力之光譜檢量模式，更有效地定量蓮霧糖度。

Independent component analysis (ICA) was integrated with near infrared (NIR) spectroscopy for rapid quantification of sugar content in wax jambu (Syzygium samarangense Merrill & Perry). The JADE algorithm (Joint Approximate Diagonalization of Eigenmatrices) and linear regression with spectral pretreatments were incorporated to analyze the NIR spectra of wax jambu against sucrose solutions. Unlike other multivariate approaches, ICA enabled comprehensive quantification of sugar content in wax jambu. In the present study, ICA was applied as the sole tool to build the NIR calibration model of internal quality of intact wax jambu without any other multivariate analysis methods. The best spectral calibration model of wax jambu (600 to 700 nm and 900 to 1,098 nm) yielded rc = 0.988, SEC = 0.243 °Brix, rv = 0.971, SEV = 0.381 °Brix, and RPD = 4.15 using the normalized first derivative spectra and 9 independent components (ICs). All ICA results were better than those of partial least squares regression (PLSR). Thus, ICA can quickly identify and effectively quantify the sugar contents in wax jambu with calibration models achieving high predictability.