VIS/NIR光譜非破壞檢測技術應用於商用木材強度性質的評估

本研究的目的在應用VIS/NIR光譜非破壞性檢測技術快速評估木材性質的特性，試材為24種商用木材，首先檢測試材的光譜特性，再測定試材的氣乾密度、抗彎強度（MOR）及壓縮強度（σc），最後分析光譜及強度性質之間的關係。以光譜掃瞄儀器測量波長範圍（350~2500 nm）反射率，透過偏最小二乘法回歸（partial least squares regression, PLSR）建立各項強度性質的預測模型，同時對所建構的PLSR模式進行預測力與精確性的評估。本研究結果顯示光譜資料經過log（1/R）轉換後，以PLSR模式對木材性質有較佳的預測能力，此表示光譜原始資料經適當的轉換，可明顯的減低光譜資料由於高頻訊號對於模式的干擾，提高模式的預測能力。研究結果顯示以光譜特性分析及評估木材物理性質的非破壞檢測技術是具有可行性的方法。

關鍵字

可見光／近紅外光光譜；木材強度性質；偏最小二乘法回歸

並列摘要

The purpose of this study was to evaluate wood physical properties by VIS/ NIR spectroscopy. The specimens of 24 commercial wood species were used for inspection. First, spectral characteristics of wood specimens were detected by spectroscopy. Then, the air-dried wood density (D), modulus of rupture (MOR), and compressive strength (σ_c) were also measured. Finally, the relationships between spectra characteristics and wood physical properties were analyzed. In this study, spectral scanning instruments to measure the wavelength range (350~2500 nm) to obtain the spectral reflectance properties. Partial Least Squares regression (PLSR) was used in order to describe the relationships between the spectral performances and wood properties. PLSR models were performed in full cross-validation mode with a maximum of several latent variables (LV). PLSR model in the wood properties had better predictive results, after spectral data by log (1/R) conversion. This represents the spectrum of the original data by the appropriate conversion, spectral data could be significantly reduced due to the high frequency signal to reduce interference pattern, improved the predictive ability of models. Our cross-validated models for D, MOR, and σ_c of the wood specimens yielded good prediction (higher the square of the product-moment correlation coefficient). Results had used spectroscopy as a rapid and reliable tool for characterizing wood products.

並列關鍵字

VIS/NIR spectroscopy ； wood physical properties ； Partial Least Squares regression

國際替代計量

VIS/NIR光譜非破壞檢測技術應用於商用木材強度性質的評估

全文下載

主題瀏覽