Partial Least Squares (PLS)法による木材の比重およぴ化学成分の推定

In order to conveniently estimate an air-dried specific gravity and a chemical abundance such as alcohol benzene extractives，holocellulose and lignin content of sugi（Cryptomeria japonica D.Don)，the partial least squares（PLS）method was done on the basis of spectrum values got from Fourier transformation Raman spectroscopy and the prediction a㏄uracy was confirmed. In the following，the results of examining prediction a㏄uracy of calibration curve in 20 unknown samples are shown，after calibration curve is made in 43 known samples.
（1） In air-dried specific gravity and holocellulose content, correlation coefficients between measured and predictive values in the known samples were 0.97 and 0.99，respectively. However，either correlation coefficients were 0.88 in the unknown samples.
（2） In the alcohol-benzene extractives content，the correlation coefficient between measured and predictive values in the known sample was 0.98，and was 0.96 in the unknown sample. However，it was indicated that attention was necessary for the applicatlon of the PLS method since the relative prediction error was about 20%.
（3） In the lignin content，the correlation coefficient between measured and predictive value in the known sample was 0.93，and was 0.92 in the unknown sample.