基于改进遗传算法的BP神经网络自适应优化设计概要

150.57140.00000.00000.00000.00000.00000.97781110 161.00000.00000.00000.00000.00000.00000.36671111

表2 测试样本

No.I nput Expect out put x1x2x3x4x5x6x7y1y2y3y4 10.00000.85710.00010.00000.00000.28570.17780000 20.00000.28570.02440.12500.50001.00000.71110001 30.14290.85710.0130.00000.00000.14290.21110010 40.00000.28570.01260.00000.00000.42860.03330011 50.58570.85710.00650.12500.00390.85710.75560100 60.57140.85710.01830.12501.00000.85710.18890101 70.28571.00000.19510.00000.00000.00000.26670110 80.85710.28570.4040.00000.00000.14290.00000111 90.14290.28570.02440.00000.00000.14290.32221000 100.28570.30360.02550.00000.00000.00000.54441001 110.57140.00000.00000.00000.00000.28570.42221010 120.28570.28570.19550.00000.00000.00000.85561011 130.71430.00000.00000.00000.00000.00000.86671100 140.28570.00000.00000.00000.00000.57140.87781101 150.57140.00000.00000.00000.00000.00000.97781110 161.00000.00000.00000.00000.00000.00000.36671111

得到网络的输入层与隐含层、隐含层与输出层最优连接权值矩阵分别如下 -1.050809.67470-6.428605.533501.12450-5.529005.123

40

5.92690-3.85980-5.96270-3.253208.451701.19850-5.26690 4.79580-7.18620-1.56390-4.52870- 5.75880-2.210900.65814-

6.47800-1.485604.296201.62720-6.81050 3.075702.68700-7.80730 -7.48030

0.618277.516801.717805.690700.10178 1.76540 3.00220 4.478604.61360 5.820504.02280 -1.84780

-4.78900 3.755102.22540 7.07340 4.71490-1.

538403.272200.917774.83660 -4.45570-1.1413013.95230

1.003104.67720

4.39130-1.79530-4.03650-1.062001.28890 4.63960 -0.933923.47560 -3.23350 -3.67760 -4.48450 -4.36950 -1.69970 -1.00240