$BB?JQNL2s5"J,@OK!$K$h$k?t

$B;38}LPO;!"7'C+!!E/!"@>2,!!MN(B


Return

$B-5(B. $B!!=o!!8@(B

$B!!8w5[<}$d7V8w$rMQ$$$?J,@O$G$O6&B8$9$kJ*@~@-$r$b$?$i$943>D$,$J$1$l$P!"%9%Z%/%H%k$KBP$9$k$=$l$>$l$N4sM?$r5a$a!"$3$l$^$G$N@~7A=E2s5"J,@OK!$K$h$jG;EY$r?dDj$G$-$k!#$7$+$7$J$,$i!"0lHLE*$K$O6&B8$9$kJ*@~@-$K1F6A$r5Z$\$9!#=E2s5"J,@O$N7gE@$r9nI~$9$k$?$a$K'$5$l$F$$$k(B1-3)$B!#$3$l$i$NJ}K!$O@bL@JQ?t$K.<+>hK!$K$h$j%b%G%k$r9=C[$9$k'$5$l$?#P#L#SK!(B1)$B$O@bL@JQ?t$K2C$($F@x:_JQ?t$rF3F~$9$k$3$H$K$h$j@bL@JQ?t$N?t$,%5%s%W%k?t$r>e2s$k>l9g$G$b8!NL%b%G%k$r9=C[$9$k$3$H$,=PMh$k!#5U9TNs1i;;$r4^$^$J$$$?$a$K6&@~@-$NLdBj$,@8$8$J$$!#@x:_JQ?t$rDL$7$F@bL@JQ?t$N>pJs$r=g4)$B!#<+F0$l$N#P#A#H$KC1N%8e7V8wJ,@O$9$kJ}K!(B5)$B$,CN$i$l!"6aG/$G$OJ,8w7V8w8!=P4oIUB0$N9bB.1UBN%/%m%^%H%0%i%UK!(B6)$BEy$,MQ$$$i$l$F$$$k!#$3$l$O#P#A#H$NB?$/$,FCM-$N7V8w$rH/$7!$$=$N7V8w%9%Z%/%H%k$OHf3SE*1T$$%T!<%/$r;}$C$F$$$k$,!"B?$/$O=E$J$C$F$$$k$N$GJ,N%A`:n$rI,MW$H$9$k$+$i$G$"$k!#$3$l$O%9%Z%/%H%k%G!]%?$N$&$A!":GBgH/8wGHD9$N$_$r;H$$!"BgItJ,$N%9%Z%/%H%k>pJs$re$OF@$i$l$?%9%Z%/%H%k%G!]%?$NA4$F$rMQ$$$kE}7W2r@O$r $B-6(B.$B!!3NG'K!!J#V#a#l#i#d#a#t#i#o#n!K(B

$B!!2s5"J,@O$NL\E*$OFHN)$KB,Dj$5$l$?%G!]%?$H%5%s%W%k$N$$$/$D$+$NB0@-$H$N4V$N4X78$r%b%G%k2=$9$k$3$H$G$"$k!#K\8&5f$G$O9g@.;nNAMO1U$N7V8w%9%Z%/%H%k$H?t $B!!!!!!!!!!(B$B!!!!!!!!!!!!!!!!!!!J#1!K(B
$B$3$3$G(B$B$O8!NL%b%G%k!"(B$B$OFHN)JQ?t$G7V8w6/EY!"$=$7$F(B$B$O=>B0JQ?t$G#P#A#H$NG;EY$G$"$k!#(B
$B!!!!!!!!!!(B$B!!!!!!!!!!!!!!!!!J#2!K(B
$B!J#2!K<0$O8!NL%b%G%k$+$iF@$i$l$?M=B,CM(B$B$H??$NCM(B$B$H$N4X78$r<($7$?$b$N$G(B$B$OM=B,8m:9$rI=$7$F$$$k!#E,Ev$J8!NL%b%G%k$rF@$k.$K$J$k(B$B$N>r7o$r5a$a$kJ}K!$G$"$k!#(B
$B!J#3!K<0$KM=B,8m:9(B$B$H#P#R#E#S#S$N4X78$r<($7$?!#(B
$B!!!!!!!!!!(B$B!!!!!!!!!!!!!!!J#3!K(B
$B$3$3$G(B$B$O8!NL%b%G%k:n@.$N$?$a$KMQ$$$?;nNAMO1U$N?t$r<($7$F$$$k!#!J#3!K<0$r0lHL2=$7$F#S#E#P!J(BStandard Error of Prediction$B!K$rDj5A$9$k!#(B
$B!!!!!!!!!!(B$B!!!!!!!!!!!!!J#4!K(B
$B8!NL%b%G%k$N0x;R?t(B$B$NA}2C$H$H$b$K#P#R#E#S#S$O8:>/$9$k$,B?$/$J$j$9$.$k$H%*!]%P!]%U%#%C%H$H$J$j!"M=B,8m:9$N860x$H$b$J$k$N$G0x;R?t$b9MN8$7$?#S#E#C!J(BStandard Error of Calibration$B!K$,Ds0F$5$l$F$$$k!#(B
$B!!!!!!!!!!(B$B!!!!!!!!!!!J#5!K(B
$B$3$N0x;R?t(B$B$O%9%3%"!"Ii2Y!"8DM-CM$+$i(BE.R.Malinowski7)$B$K$h$k#I#D#NK!$d#F%F%9%H$K$h$C$F7W;;$5$l$k$,7P83E*$KN> $B!!#S#E#P$d#S#E#C$K$h$C$FF@$i$l$?CM$O8!NL%b%G%k:n@.$KMQ$$$?%G!<%?%;%C%H$K$D$$$F7W;;$7$FF@$i$l$?CM$G!"8!NL%b%G%k$H$OJL$N%G!<%?%;%C%H$K$D$$$F8!F$$7$?$b$N$G$O$J$$!#=>$C$F8!NL%b%G%k$NM-8z@-$Ko$KMQ0U$G$-$k$H$O8B$i$J$$!#$=$N2r7h:v$H$7$F%/%m%9%P%j%G!<%7%g%sK!$,Ds0F$5$l$F$$$k!#$3$l$O%G!<%?%;%C%HCf$N%5%s%W%k$r#1$D=|$$$?8!NL%b%G%k$r:n@.$7!"=|$$$?%5%s%W%k$NM=B,$r9T$$8m:9$r5-O?$9$k!#$3$N$B#S#E#C#V(B(Standard Error of Cross-validation)$B$HDj5A$9$k!#$3$l$O8!NL%b%G%kCf$K$J$$%5%s%W%k$rMQ$$$F7W;;$7$?$3$H$K$J$k$N$G$h$j?.Mj@-$,9b$$$H9M$($i$l$F$$$k!#K\8&5f$G$O#5$l#S#E#P!"#S#E#C$*$h$S#S#E#C#V$r7W;;$7$=$NJ?6QCM!J(B$B!$(B$B!$(B$B!K$r8!NL%b%G%k:n@.$NL\0B$H$7$?!#>0!"#P#L#SK!$N?t3XE*GX7J$K$D$$$F$OB??t$NO@J8(B8-12)$B$d@.=q(B13-17)$B$,=PHG$5$l$F$$$k$N$G;2>H$5$l$?$$!#(B

$B-7(B.$B!!

$B-7!%#1!!;n!!Lt(B

$B!!OB8w=cLt@=$N%Y%s%>(B[$B#a(B]$B%T%l%s(B(B[a]p)$B!"%Z%j%l%s(B(Pery)$B!"%U%k%*%i%s%F%s(B(Fluo)$B!"%T%l%s(B(Py)$B$*$h$S%/%j%;%s(B(Chry)$B$NLs#1#0-S$r@:Gi$7!"L57V8w@-$N#n!]%X%-%5%s(B100ml$B$KMO2r$7$FJ]B8MO1U$H$7!"Nd0E=j$KJ]B8$7$?!#;HMQ$K:]$7$F$O$5$i$K#1&L(Bg$B!?(Bml$B$K4u $B-7!%#2!!9g@.;nNAMO1U(B

$B!!(B $B!!#1#0(Bml$B$N%a%9%U%i%9%3$rMQ$$$F%Y%s%>(B[a]$B%T%l%s!"%Z%j%l%s!"%U%k%*%i%s%F%s!"%T%l%s$*$h$S%/%j%;%s$NG;EY$r(B0.01$B!A(B0.1$B&L(Bg$B!?(Bml$B$NHO0O$G $B-7!%#3!!B,DjAuCV(B Table 1 Experimental conditions of fluorometry
Ex$B&K!JNe5/GHD9!K(B $B#3#0#5#n#m(B
Em$B&K!J7V8wGHD9!K(B $B#3#3#0!A#5#2#0#n#m(B
$B#P#M#T!!#G#a#i#n(B Medium
$BNe5/%P%s%II}(B $B#1#0#n#m(B
$B7V8wB&%P%s%II}(B $B#1#0#n#m(B
$B%G!<%?$N $B#1(B.$B#0#n#m(B

$B!!D4@=$7$?;nNAMO1U$N7V8w%9%Z%/%H%k$NB,Dj$K$O%9%Z%/%H%kJd@5$r$7$?F|K\J,8w9)6Hr7o$r(BTable 1$B$K<($9!#(B

$B-7!%#4(B $B%H%l!<%K%s%0%;%C%H(B

$B!!-7!%#2$GD4@=$7$?#2#9Table 1$B$N>r7o$GB,Dj$7!"%V%i%s%/Jd@58e#A#S#C#I#I7A<0$KJQ49$7$F%m!]%?%9$KFI$_9~$`!#7V8w%9%Z%/%H%kIt#1#9#1%+%i%`$K#5$lFHN)JQ?t$*$h$S=>B0JQ?t$H$7$FDj5A$7$?!#$3$N%G!]%?%^%H%j%C%/%9$r4p$K$7$F;nNA?t!"%G!]%?r7o$rJQ$($F9=C[$7$?%H%l!]%K%s%0%;%C%H$r(BTable 2$B$K<($7$?!#$3$N$&$A;nNAHV9f#1!A#2#1$r8!NL%b%G%k:n@.$KMQ$$!";nNAHV9f#2#2!A#2#9$r8!NL%b%G%k$NM-8z@-$NH=Dj$KMQ$$$?!#(B

$B!!!!(BTable 2 Training sets
No. Training
Set
Number$B!!(Bof
Samples
Interval
(nm)
Independent
Variables
Dependent
Variables
1 D1 29 1 $B!!(B191 5
2 D2 29 2 $B!!(B96 5
3 D4 29 4 $B!!(B48 5
4 D8 29 8 $B!!(B24 5
5 D20 29 20 $B!!(B10 5
6 D30 29 30 $B!!(B7 5

$B-7!%#5!!%G!]%?2r@O(B

$B!!%G!<%?$N#P#L#S#1(B10)$B2r@O$K$D$$$F$O!"%G!<%?$rCf1{J?6Q2=(B(Mean Centering)$B$7$?8e(BInfoMetrix$Be7?%3%s%T%e!<%?>e$G$*$3$J$C$?!#(B

$B-8!%!!7k2L$H9M;!(B

$B-8!%#1!!7V8w%9%Z%/%H%k(B

$B#5Table 1$B$N>r7o$GB,Dj$7$?!#7k2L$r(BFig.1$B$N(B(1)$B!A(B(5)$B$K<($7$?!#$=$l$>$l$N#P#A#H$OFCD'$N$"$k%9%Z%/%H%k$r;}$D$,=E$J$C$F$$$k!#DL>o$3$l$i$N#P#A#H$rDjNL$9$k>l9g!"GvAj%/%m%^%H%0%i%UK!$d%,%9%/%m%^%H%0%i%UK!$K$h$j$"$i$+$8$aJ,N%8e8DJL$KDjNL$9$kJ}K!$,:NMQ$5$l$F$$$k!#%9%Z%/%H%k(B(6)$B$O%H%l!<%K%s%0%;%C%H$KMQ$$$?9g@.;nNAMO1U$N7V8w%9%Z%/%H%k$N0lNc$G$"$k!#6&B8$7$F$$$k#5D$7$F$$$k$N$G#5 $B-8!%#2!!;nNA?t$H(B$B!$(B $B!!NI9%$J8!NL%b%G%k$r:n@.$9$k>e$G%H%l!<%K%s%0%;%C%H$NBg$-$5$,LdBj$K$J$k!#%H%l!<%K%s%0%;%C%H$,>.$5$$$HL$CN$N;nNAMO1U$NFCD'$r==J,@bL@$G$-$J$$$3$H$OL@Gr$G$"$k!#0lJ}(B


Fig.1 Fluorescence spectra of PAHs.
(1) B[a]p:10-1$B&L(Bgcm-3; (2) Pery:10-1$B&L(Bgcm-3; (3) Fluo:10-1$B&L(Bgcm-3;
(4) Py:10-1$B&L(Bgcm-3; (5) Chry:10-1$B&L(Bgcm-3;(6) B[a]p:9$B!_(B10-2$B&L(Bgcm-3,
Pery:10-2$B&L(Bgcm-3, Fluo:9$B!_(B10-2$B&L(Bgcm-3,Py:4$B!_(B 10-2$B&L(Bgcm-3,Chry:7$B!_(B10-2$B&L(Bgcm-3; Solvent: n-Hexane; Ex$B&K(B=305nm,Em$B&K(B=330$B!A(B520nm.


Fig.2 Relationship between and the sample number of training set.
(1):D1,(2):D2,(3):D4,(4)D30

$BBg$-$9$.$k%H%l!<%K%s%0%;%C%H$OKDBg$J7W;;$K$h$k8m:9$N=8@Q$,M=B,@:EY$K1F6A$9$k$3$H$b9MN8$KF~$l$J$1$l$P$J$i$J$$!#$=$3$G(BTable 2$B$NA4$F$N%H%l!<%K%s%0%;%C%H$rMQ$$$F8!NL%b%G%k$KMQ$$$i$l$k;nNA$N?t$,#S#E#C$H#S#E#C#V$K$*$h$\$91F6A$rD4$Y$?!#(BFig.2$B$*$h$S(BFig.3$B$K7k2L$N0lIt$r<($7$?!#2#<4$O%H%l!<%K%s%0%;%C%H$r9=@.$9$k$?$a$KMQ$$$?;nNAMO1U$N?t$G$"$k!#=D<4$O%H%l!<%K%s%0%;%C%H$rMQ$$$F#5$l$N#P#R#E#S#S$r#S#E#C$*$h$S#S#E#C#V$K49;;$7$?8eJ?6Q$7$?$b$N$G$"$k!#?^Cf$N!J#1!K!"!J#2!K!"!J#3!K$*$h$S!J#4!K$O$=$l$>$l#D#1!"#D#2!"#D#4$*$h$S#D#3#0$rI=$7$F$$$k!#8!F$$7$?%H%l!<%K%s%0%;%C%H$O2?$l$b$h$/(B


Fig.3 Relationship between and the sample number of training set.
(1):D1,(2):D2,(3):D4,(4):D30

$B;w$?798~$r<($9$,#D#3#0$N>l9g!"9b$$#S#E#CCM$r<($7$F$$$k!#$3$l$O#5(B[a]$B%T%l%s!"%Z%j%l%s$*$h$S%U%k%*%i%s%F%s$N#P#R#E#S#S$,9b$$$3$H$,860x$G$"$C$?!#$3$l$O$3$l$i$N#P#A#H$NFCD'$r<($97V8w$r==J,$KJa$i$($-$l$J$+$C$?$?$a$G$"$k!#%H%l!<%K%s%0%;%C%H$NFb#D#1!A#D#8$O$[$\F1$8798~$r<($7!"#D#2#0$O$3$l$i$H#D#3#0$NCf4VE*$J798~$r<($7$?!#8!F$$7$?HO0O$G(B$BCM$O;nNA?t$NA}2C$KH<$C$F6O$+$G$"$k$,8:>/798~$K$"$k$,(B$BCM$O;nNA?t$,#1#20J>e$G$[$\0lDj$NCM$r<($7$F$$$k!#(B

$B-8!%#3!!8!NL%b%G%k(B

$B!!-8!%#2$N8!F$$G#D#1!A#D#8$N%H%l!<%K%s%0%;%C%H$rMQ$$$F9=C[$5$l$k8!NL%b%G%k$O#S#E#C$*$h$S#S#E#C#V$N2?$l$N>r7o$K$*$$$F$b%H%l!<%K%s%0%;%C%H$KMQ$$$i$l$k;nNA?t$,#1#20J>e$GF1DxEY$NM=B,@-G=$r;}$D$3$H$,M=A[$5$l$k$,#S#E#C$*$h$S#S#E#C#V4V$NM=B,@-G=$NHf3S$O$G$-$J$+$C$?!#$=$3$G(BTable 2 $B$N%H%l!<%K%s%0%;%C%H$G#S#E#C$*$h$S#S#E#C#V$N8!NL%b%G%k$N9=C[$KMQ$$$i$l$J$+$C$?#48D$N%G!<%?$rA*$S!"#5Fig.2$B$*$h$S(BFig.3$B$N>r7o$G$=$l$>$l9=C[$7$?8!NL%b%G%k$rMQ$$$F#5Fig.4$B$O#D#4$K$D$$$F$N8!F$7k2L$r<($7$?$b$N$G$"$k!#(B(1)$B$*$h$S!J(B2$B!K$O$=$l$>$l(BFig.2$B$*$h$S(BFig.3$B$K$*$1(B


Fig.4 Comparison of calibration model.
Trainingset:D4,(1):,(2):(SEC), (3): ,(4): (SECV)

$B$k#D#4$N(B$B$*$h$S(B$B$G$"$k!#$^$?(B(3)$B$*$h$S(B(4)$B$O$=$l$>$l#S#E#C$*$h$S#S#E#C#V$KBP1~$9$k8!NL%b%G%k$K$h$k(B$B$r<($7$F$$$k!#7k2L$O(B$B$H#S#E#C#V$K$h$k8!NL%b%G%k$K$h$k(B(SECV) $B$,8!F$$7$?%H%l!<%K%s%0%;%C%H$N;nNA?t$NHO0O$G6K$a$F$h$/;w$?798~$r<($7!"#S#E#C$N>l9g$HHf3S$7$F$b#S#E#C#V$NM=B,@-G=$,9b$$$3$H$r<($7$F$$$k!#$3$l$O8!F$$7$?#D#1!"#D#2!"#D#8$K$D$$$F$b$[$\F1MM$N798~$G$"$C$?!#(B

$B-8!%#4!!M=B,@-G=(B

$B!!9g@.$7$?;nNAMO1U$NFb!"8!NL%b%G%k$KMQ$$$J$+$C$?#4;nNAMO1U$h$jG;EY9`$r=|$$$?%G!<%?$K-8!%#3$G9=C[$7$?#S#E#C#V$N>r7o$K$h$k8!NL%b%G%k$rE,MQ$7!"M=B,@-G=$r8!F$$7$?!#7k2L$O#D#1!A#D#4$K$D$$$F!"%b%G%k$K;HMQ$7$?;nNA?t$,#1#2!A#2#0$G$OF1DxEY$NNI9%$JM=B,@-G=$r<($7!";nNA?t$NA}2C$K$h$k1F6A$O8+$i$l$J$+$C$?!#(BTable 3 $B$O%H%l!<%K%s%0%;%C%H#D#4!"%5%s%W%k?t#1#4$N8!NL%b%G%k!J#L#v#M#1#4#D#4!K$rMQ$$$?>l9g$NM=B,CM$*$h$S$l!\(B0.39$B!J!_(B10-2$B&L(Bgcm-3 $B!$(BChry)$B$*$h$S(B0.252$B!J!_(B10-2$B&L(Bgcm-3 $B!$(BChry)$B$G$"$C$?!#(B $B0J>e$N$3$H$+$i!"6&B8$9$k?tr7o$r8!F$$7$F$$$-$?$$$H9M$($F$$$k!#$^$?!"$3$l$^$G6&B8J*$NJ,@OJ}K!$K$D$$$F$bK\K!$NE,MQ$r8!F$$7$F$$$-$?$$!#(B

Table 3 Simultaneous five components analysis of PAHs by means of PLS$B!!(B
Sample no. B[a]p Pery Fluo Py Chry
1 Actual conc.
PLS, (10-2$B&L-Q(B-3)
4.00
3.97
7.00
6.82
7.00
6.80
10.00
9.74
10.00
9.72
2 Actual conc
PLS, (10-2$B&L-Q(B-3)
2.00
1.98
10.00
10.24
6.00
6.13
5.00
5.02
4.00
3.87
3 Actual conc.
PLS, (10-2$B&L-Q(B-3)
8.00
8.20
8.00
8.22
8.00
8.14
6.00
6.00
3.00
2.92
4 Actual conc.
PLS, (10-2$B&L-Q(B-3)
2.00
1.92
3.00
2.84
2.00
1.87
9.00
9.18
9.00
9.39
SEP(10-2$B&L-Q(B-3) 0.109 0.202 0.156 0.157 0.252
Average SEP of five components:0.175($B!_(B10-2$B&L-Q(B-3$B!K(B; Model(LvM14D4)

$B-:(B.$B!!J8!!8%(B

1) W.J.Dunn,III, S.Wold, J.Edlund, S.Hellberg, and J.Gasteiger, Quant.Struct.Act. Relat.,3,131 (1984).
2) K.R.Beebe, and B.R.Kowalski, Anal.Chem. 59,1007A (1987).
3) H.Martens and T.Naes, "Multivariate Calibration", John Wiley & Sons, p.98 (1989).
4) $B5\2!(B, $B:4!9LZ?50l(B, "$B%1%b%a%H%j%C%/%9!]2=3X%Q%?!<%sG'<1$HB?JQNL2r@O!](B", $B6&N)=PHG(B, p. 55-56 (1995).
5) $B>>2<=(Da(B, $BBg5$1x@w8&5f(B, 10,723 (1976).
6) K.Ogan, E.Katz, W.Skavin, Anal. Chem., 51,1315 (1979).
7) E.R.Malinowski,"Factor Analysis in Chemistry", 2nd ed., John Wiley & Sons, p.98 (1991).
8) P.Geladi, and B.R.Kowalski, Anal.Chim. Acta., 185,1 (1986).
9) A.Lober, L.E.Wangen, and B.R.Kowalski, J.Chemom., 1,19 (1987).
10) R.Manne, Chemom.Intell.Lab.Syst. 2,187 (1987).
11) D.M.Haaland, and E.V.Thomas,"Partial Least-Squares Method for Spectral Analysis. 1. Relation to Other Quantitative Calibration Methods and the Extraction of Qualitative Information", Anal.Chem., 60,1193 (1988).
12) D.W.Osten, "Selection of Optimal Regression Models via Cross-Validation", J.Chemometrics, 2,39 (1988).
13) W.H.Press, B.P.Flanny, S.A.Teukolsky, and W.T.Vetterling,"Numerical Recipes", Cambridge University Press, p.52 (1986).
14) H.Wold, "Partial Least Squares", in S.Katz and N.L.Johnson, ed., Encyclopedia of Statistic Sciences, Vol.6, Wiley,New York, p.581 (1985).
15) S.Weisberg, "Applied Linear Regression",Chapter 5, John Wiley & Sons, New York (1982).
16) R.D.Cook, and S.Weisverg, "Residuals and Influence in Regression", Chapter2 John Wiley & Sons, New York (1982).
17) $BAjEgoDO:!$%1%b%a%H%j%C%/%9!$(B3,4,7$B>O!$4]A1(B (1992).

Return