做响应面的问题,高手指教
我在实验中用Box-Behnken做A、B、C三因素的交互作用并做响应面分析,利用Design-Expert软件得到如下方差分析结果:
Sum of Mean F p-value
Source Squares df Square Value Prob > F
Model 806.37 9 89.60 16.39 0.0006 significant
A-A 81.92 1 81.92 14.99 0.0061
B-B 0.18 1 0.18 0.033 0.8611
C-C 426.32 1 426.32 78.01 < 0.0001
AB 0.20 1 0.20 0.037 0.8528
AC 18.06 1 18.06 3.31 0.1119
BC 97.02 1 97.02 17.75 0.0040
A2 99.55 1 99.55 18.22 0.0037
B2 5.69 1 5.69 1.04 0.3415
C2 62.82 1 62.82 11.49 0.0116
Residual 38.26 7 5.47
Lack of Fit 36.66 3 12.22 30.55 0.0032 significant
Pure Error 1.60 4 0.40
Cor Total 844.62 16
因为有些项不显著,而且失拟项极显著,因而我剔除了B-B、AB两项,并把A2C加入方程,结果如下:
Sum of Mean F p-value
Source Squares df Square Value Prob > F
Model 837.19 8 104.65 112.64 < 0.0001 significant
A-A 81.92 1 81.92 88.17 < 0.0001
C-C 113.42 1 113.42 122.08 < 0.0001
AC 18.06 1 18.06 19.44 0.0023
BC 97.02 1 97.02 104.43 < 0.0001
A2 99.55 1 99.55 107.15 < 0.0001
B2 5.69 1 5.69 6.12 0.0384
C2 62.82 1 62.82 67.61 < 0.0001
A2C 31.21 1 31.21 33.59 0.0004
Residual 7.43 8 0.93
Lack of Fit 5.83 4 1.46 3.65 0.1191 not significant
Pure Error 1.60 4 0.40
Cor Total 844.62 16
不知道这样做是否正确?
Sum of Mean F p-value
Source Squares df Square Value Prob > F
Model 806.37 9 89.60 16.39 0.0006 significant
A-A 81.92 1 81.92 14.99 0.0061
B-B 0.18 1 0.18 0.033 0.8611
C-C 426.32 1 426.32 78.01 < 0.0001
AB 0.20 1 0.20 0.037 0.8528
AC 18.06 1 18.06 3.31 0.1119
BC 97.02 1 97.02 17.75 0.0040
A2 99.55 1 99.55 18.22 0.0037
B2 5.69 1 5.69 1.04 0.3415
C2 62.82 1 62.82 11.49 0.0116
Residual 38.26 7 5.47
Lack of Fit 36.66 3 12.22 30.55 0.0032 significant
Pure Error 1.60 4 0.40
Cor Total 844.62 16
因为有些项不显著,而且失拟项极显著,因而我剔除了B-B、AB两项,并把A2C加入方程,结果如下:
Sum of Mean F p-value
Source Squares df Square Value Prob > F
Model 837.19 8 104.65 112.64 < 0.0001 significant
A-A 81.92 1 81.92 88.17 < 0.0001
C-C 113.42 1 113.42 122.08 < 0.0001
AC 18.06 1 18.06 19.44 0.0023
BC 97.02 1 97.02 104.43 < 0.0001
A2 99.55 1 99.55 107.15 < 0.0001
B2 5.69 1 5.69 6.12 0.0384
C2 62.82 1 62.82 67.61 < 0.0001
A2C 31.21 1 31.21 33.59 0.0004
Residual 7.43 8 0.93
Lack of Fit 5.83 4 1.46 3.65 0.1191 not significant
Pure Error 1.60 4 0.40
Cor Total 844.62 16
不知道这样做是否正确?
没有找到相关结果
已邀请:
1 个回复
renjiescut13 (威望:0) (广东 广州)
赞同来自:
从失拟显著来看,你试验的误差比较大,检测方法可能平行之间的重复性不好。
其它还不好说,希望有高手指点。