| 摘要： 本研究乃利用近紅外光線光譜分析之非破壞性檢測方法針對特定水果－「番石榴」進行糖度的檢測，以波長400nm~2498nm為檢測範圍。以屏東縣內門鄉所產之番石榴為樣本，並以糖度曲折度計分析其可溶性固形物，以建立其糖度之校正方程式，利用近紅外線光譜分析達成快速且非破壞檢測番石榴糖度，做為線上檢測分級之參考。 實驗結果顯示有關番石榴糖度之近紅外線光譜以二次微分光譜分析結果較佳。在數值分析上，則以MPLSR模式的分析結果比MLR模式好。在MPLSR模式中，因子數為10時，總校正線之RSQ值為0.941，SEC值為0.309。而在PLSR模式中，因子數為12時，總校正線之RSQ值為0.898，SEC值為0.417。而在MLR模式中，於選用波長數為7個時，總校正線之RSQ值為0.846，SEC值為0.537，呈現出較佳的預測結果。研究結果可發現利用近紅外線光譜進行番石榴糖度分析確實可行，同時也發現在MPLS、MLR、PLS三種迴歸模式中，以MPLS迴歸模式所建立之校正方程式最為具實用性。This research study used near infrared spectrum to detect the sugar content of guava with nondestructive detection method, given the wavelength range of 400nm~2498nm. The sugar content spectrometer was used to analyze the soluble solid substance in guava produced in Neimen Shiang, Pingtung, and establish the sugar content calibration equation. The near infrared spectrum was employed to analyze the guava sugar content with quick and non-destructive detection for online detection rating. The results showed that the near infrared spectrum of guava sugar content was best measured with quadratic differential spectrometer. In the numerical analysis, the MPLSR model exhibited more desirable results than MLR model. In MPLSR model, when the factor was 10, RSQ of the total calibration line was 0.941, and SEC was 0.309. In PLSR model, when the factor was 12, RSQ of the total calibration line was 0.898, and SEC was 0.417. In the MLR model, when the number of wavelength used was 7, RSQ of the total calibration line was 0.846, and SEC was 0.537, which was more desirable. The results proved the feasibility of the use of near infrared spectrum on guava sugar content detection, and discovered that the calibration equation established based on the MPLS regression model was most practical among MPLSR, MLR, and PLSR models. |
頁面載入中請稍後...