试验设计及数据挖掘技术课程总练习完美修订版.docx

1、试验设计及数据挖掘技术课程总练习完美修订版课程总练习一、试验设计全面试验法单因素多次试验法正交设计法均匀设计法配方均匀设计法调优均匀设计法三次均匀设计法二、数据挖掘技术2.1逐步回归分析建模线性模型二次模型高次模型混合模型2.2最优化计算寻优网格优法法蒙特卡罗优化法均匀设计优化法数论序贯优化法2.3交互作用Y值等高线图分析三、验证试验1、均匀设计试验方案的构造已知一试验有四个因素，他们的试验范围及因素水平如下：X1:112；X2：200350 ；X3: 1020；X4:20004000 表1、因素水平表NO.123456789101112X1123456789101112X2200250300

2、350X3101214161820X420002200240026002800300032003400360038004000（1）给出12拟水平的因素水平表参考答案表1B、因素拟水平表NO.123456789101112X1123456789101112X2200200200250250250300300300350350350X3101012121414161618182020X4200022002400260028003000320034003600380040002000（2）给出12拟水平的试验方案参考答案表1C、U12（124）试验方案表NO.1234567891011121123

3、456789101112X11234567891011126612511410392817X22503502503502503502003002003002003007831161941272105X31612201410181220161018149107411185212963X43800320026002000400034002800220020003600300024002、回归分析建模请对表2的数据进行逐步回归分析表2、试验方案及结果No.X1X2X3X4X5Y11540357016053822060651101504563251002050140472430205011013042

4、053540805012061464080209016055774510050301506828502080901405629556035301305601060806570120550（1）、用一次模型，要求数据中心化处理，寻找出最好的回归方程。 模型： Y= A0 + A1X1 + A2X2 + A3X3+ A4X4+ A5X5 回归分析结果: I=1 B=1.97557894736842 F=3.67396503815582I=3.0359* F=8.04474411017923I=4 B=-2.05065789473684.024*I=5 B=3.43498245614035.0875

5、*BO=541.1 F=7.57031378775752 R=.926434933430933 S=38.7528855745076 FO=1.54 回归方程: Y*= 541.1+( 1.97557894736842)*(X1- 37.5)+( 2.03590643274854)*(X3- 50)+(-2.05065789473684)*(X4- 70)+( 3.43498245614035)*(X5- 140)-（2）、用一次模型加上交叉项，要求数据中心化处理，寻找出最好的回归方程, 并且剔除不显著项从而寻找出最好的回归方程。模型：Y= A0 + A1X1 + A2X2 + A3X3+ A

6、4X4+ A5X5+ A6X1X2+ A7X1X3+ A8X1X4+ A9X1X5 + A10X2X3+ A11X2X4+ A12X2X5+ A13X3X4+ A14X3X5+ A15X4X5 回归分析结果: I=1 B=1.97557894736842 F=162.670156787581I=3.0359* F=356.192770515776I=4 B=-2.05065789473684 F=753.797654882556I=5 B=3.43498245614035 F=402.362944012816I=13 B=-6.04761904761905E-02 F=217.38228739

7、0022BO=548.357142857143 F=311.625798693041 R=.998718875359091 S=5.8239496670483 FO=1.54- 回归方程: Y*= 548.357142857143+( 1.97557894736842)*(X1- 37.5)+( 2.03590643274854)*(X3- 50)+(-2.05065789473684)*(X4- 70)+( 3.43498245614035)*(X5- 140)+(-6.04761904761905E-02)*(X3- 50)*(X4- 70)-（3）、用二次模型，要求数据中心化处理，寻找出

8、最好的回归方程。模型：Y= A0 + A1X1 + A2X2 + A3X3+ A4X4+ A5X5+ A6X12+ A7X22+ A8X32+ A9X42 + A10X52- 回归分析结果: I=1 B=1.97557894736842 F=6.178*97I=3.0359* F=13.5277381511204I=4 B=-2.05065789473684 F=28.6282545246892I=5 B=3.43498245614035 F=15.281221290476I=10 B=.118571428571429 F=4.40781133983083BO=517.385714285714

9、 F=11.0655254856918 R=.965700670950976 S=29.8846083845136 FO=1.54 回归方程: Y*= 517.385714285714+( 1.97557894736842)*(X1- 37.5)+( 2.03590643274854)*(X3- 50)+(-2.05065789473684)*(X4- 70)+( 3.43498245614035)*(X5- 140)+( .118571428571429)*(X5- 140)2-（4）、用二次模型加上交叉项，要求数据中心化处理，寻找出最好的回归方程。模型： Y= A0 + A1X1 + A2

10、X2 + A3X3+ A4X4+ A5X5+ A6X12+ A7X22+ A8X32+ A9X42 + A10X52+ A11X1X2+ A12X1X3+ A13X1X4+ A14X1X5 + A15X2X3+ A16X2X4+ A17X2X5+ A18X3X4+ A19X3X5+ A20X4X5- 回归分析结果: I=1 B=1.97557894736842 F=162.670156787581I=3.0359* F=356.192770515776I=4 B=-2.05065789473684 F=753.797654882556I=5 B=3.43498245614035 F=402.3

11、62944012816I=13 B=-6.04761904761905E-02 F=217.382287390022BO=548.357142857143 F=311.625798693041 R=.998718875359091 S=5.8239496670483 FO=1.54 回归方程: Y*= 548.357142857143+( 1.97557894736842)*(X1- 37.5)+( 2.03590643274854)*(X3- 50)+(-2.05065789473684)*(X4- 70)+( 3.43498245614035)*(X5- 140)+(-6.04761904

12、761905E-02)*(X3- 50)*(X4- 70)（5）、用二次模型加上交叉项，要求数据中心化处理，并且剔除不显著项从而寻找出最好的回归方程。模型： Y= A0 + A1X1 + A2X2 + A3X3+ A4X4+ A5X5+ A6X12+ A7X22+ A8X32+ A9X42 + A10X52+ A6X1X2+ A7X1X3+ A8X1X4+ A9X1X5+ A10X2X3+ A11X2X4+ A12X2X5+ A13X3X4+ A14X3X5+ A15X4X5- 回归分析结果: I=1 B=1.97557894736842 F=162.670156787581I=3.0359*

13、 F=356.192770515776I=4 B=-2.05065789473684 F=753.797654882556I=5 B=3.43498245614035 F=402.362944012816I=13 B=-6.04761904761905E-02 F=217.382287390022BO=548.357142857143 F=311.625798693041 R=.998718875359091 S=5.8239496670483 FO=1.54 回归方程: Y*= 548.357142857143+( 1.97557894736842)*(X1- 37.5)+( 2.03590

14、643274854)*(X3- 50)+(-2.05065789473684)*(X4- 70)+( 3.43498245614035)*(X5- 140)+(-6.04761904761905E-02)*(X3- 50)*(X4- 70)-II、对方程各项进行F检验f1=1, f2=8, F0.01=11.3, F=162 F0.01=11.3，各项通过a=0.01III、对整个方程（或者总方程）进行F检验f1=6-1=5, f2=10-5-1=4, F0.01=15.5, F=311.6 F0.01=15.5,总方程通过.3、最优化计算（1）用网格优化法对下列的方程进行单指标优化计算求出最

15、大值及其对应的参数，并列出原程序Y = 34.9272 + 1.178510-3*(X2-120)2 + 9.533810-2*(X1-6.2)*(X2-120) + 5.004510-3*(X2-120)*(X3-30)X1：5.07.4；X2：40200；X3：2040，实验中Y的最大值为38.09Option ExplicitPublic Sub YHLX1()10 Dim G As Integer, J As Integer20 Dim ZM, S1,S2,S3, X1,X2, X3, Y, Y1, Y230 Dim WS As Worksheet, WF As WorksheetFu

16、nction40 Set WS = Application.ActiveSheet: Set WF = Application.WorksheetFunction50 ZM = WS.Cells(8, 2): G = WS.Cells(8, 4)60 S1 = (7.4 -5.0) / G: S2 = (200-40) / G :S3 = (40 - 20) / G70 J = 180 For X1 = 5.0 To 7.4 + S1 / 2 Step S185 For X2 = 40 To 200 + S2 / 2 Step S290 For X3 = 20 To 40 + S3 / 2 S

17、tep S3100 Y1 = 34.9272 + 1.1785e-3*(X2-120)2110 Y2 = 9.5338e-2*(X1-6.2)*(X2-120) + 5.0045e-3*(X2-120)*(X3-30)120 Y = Y1 + Y2130 If Y ZM Then ZM = Y150 WS.Cells(8, 7) = The Results of Optimization160 WS.Cells(8 + J, 6) = X1=: WS.Cells(8 + J, 7) = X1165 WS.Cells(8 + J, 8) = X2=: WS.Cells(8 + J, 9) = X

18、2170 WS.Cells(8 + J, 10) = X3=: WS.Cells(8 + J, 11) = X3180 WS.Cells(8 + J, 12) = Y=: WS.Cells(8 + J, 13) = Y190 J = J + 1200 Next X3,X2, X1210 End End SubX1=5X2=40X3=20Y=55.62565X1=5X2=40X3=20.95238Y=55.24435X1=5X2=40X3=21.90476Y=54.86306X1=5X2=40X3=22.85714Y=54.48176X1=5X2=40X3=23.80952Y=54.10047X

19、1=5.114286X2=40X3=20Y=54.75399X1=5.114286X2=40X3=20.95238Y=54.37269X1=7.285714X2=200X3=39.04762Y=54.37269X1=7.285714X2=200X3=40Y=54.75399X1=7.4X2=200X3=36.19048Y=54.10047X1=7.4X2=200X3=37.14286Y=54.48176X1=7.4X2=200X3=38.09524Y=54.86306X1=7.4X2=200X3=39.04762Y=55.24435X1=7.4X2=200X3=40Y=55.62565（2）用

20、蒙特卡罗法对下列的方程进行单指标优化求出最逼近理论值的最优参数，并列出原程序Y = 34.9272 + 1.178510-3*(X2-120)2 + 9.533810-2*(X1-6.2)*(X2-120) + 5.004510-3*(X2-120)*(X3-30)X1：5.07.4；X2：40200；X3：2040，实验中Y的最大值为38.09Public Sub YHLX2()10 Dim I As Long20 Dim ZM, S1,S2, S3, X1,X2,X3, Y, Y1, Y2, A1, A330 Dim G As Long, N As Long40 Dim WS As Wor

21、ksheet, WF As WorksheetFunction50 Set WS = Application.ActiveSheet: Set WF = Application.WorksheetFunction60 ZM = WS.Cells(8, 2): N = WS.Cells(8, 4): G = 10 * N70 S1 = (7.4 -5.0) / N: S2 = (200 - 40) / N: S3 = (40 - 20) / N80 J = 190 For I = 1 To G100 A1 = Int(N * Rnd(1)105 A2 = Int(N * Rnd(1)110 A3

22、 = Int(N * Rnd(1)120 X1 = 5 + A1 * S1125 X2 = 40 + A2 * S2130 X3 = 20 + A3 * S3140 Y1 = 34.9272 + 1.1785E-3*(X2-120)2 + 9.5338E-2*(X1-6.2)*(X2-120)150 Y2 = 5.0045E-3*(X2-120)*(X3-30)160 Y = Y1 + Y2170 If Y ZM Then ZM = Y190 WS.Cells(8, 7) = The Results of Optimization200 WS.Cells(8 + J, 6) = X1=: WS

23、.Cells(8 + J, 7) = X1205 WS.Cells(8 + J,8) = X2=: WS.Cells(8 + J,9) = X2210 WS.Cells(8 + J,10) = X3=: WS.Cells(8 + J, 11) = X3220 WS.Cells(8 + J, 12) = Y=: WS.Cells(8 + J, 13) =Y230 J = J + 1240 Next I250 EndEnd SubX1=5.047232X2=40.352X3=20.0408Y=55.12662X1=7.398512X2=199.2736X3=39.5028Y=55.16135X1=

24、7.390592X2=199.6096X3=39.8456Y=55.35509X1=5.033888X2=40.0416X3=20.0252Y=55.34255X1=5.013632X2=40.0192X3=20.3096Y=55.391X1=5.021648X2=40.928X3=20.278Y=55.02587X1=5.03288X2=40.0832X3=20.068Y=55.31857X1=5.025968X2=40.3744X3=20.368Y=55.14987X1=7.379504X2=199.9424X3=39.1788Y=55.120574、配方均匀设计（1）一个饲料的配方由四种

25、主要的成分组成，根据试验条件的允许和精度的要求，需要选择UM21（214）表来安排试验，请用相应的软件生成该配方试验方案表。I、用U21生成UM21（214）No.X1X2X3X410.7123152090.0941900980.0506770.14281820.5850867330.0150928390.199910.1999130.5080659270.1789611830.2310040.08196940.4496787920.0337948760.5042280.01229850.401592 0.240791 0.0766320.28098560.360195 0.055746 0.2642170.31984270.323557 0