24小时热门版块排行榜

返回列表

当前只显示满足指定条件的回帖，点击这里查看本话题的所有回帖

yuweibest

木虫 (正式写手)

应助: 2 (幼儿园)
金币: 2455.2
散金: 12
红花: 1
帖子: 988
在线: 127.5小时
虫号: 704970
注册: 2009-02-20
性别: GG
专业: 微生物资源与分类学

[交流] 【求助/交流】神经网络遗传算法优化培养基已有1人参与

本人在用神经网络遗传算法优化培养基时老是出现问题，现在向各位大虾求助，帮忙用matlab编下程序，在下感激不尽
P=[0.59920,0.59920,0.59920,0.59920,0.71944,0.71944,0.71944,0.71944,0.79959,0.79959,0.79959,0.79959,1,1,1,1;
0.07816,0.09819,0.11824,0.13828,0.07816,0.09819,0.11824,0.13828,0.07816,0.09819,0.11824,0.13828,0.07816,0.09819,0.11824,0.13828;
0.05811,0.07816,0.09819,0.118240,07816,0.05811,0.11824,0.09819,0.09819,0.11824,0.05811,0.07816,0.11824,0.09819,0.07816,0.05811;
0,0.00200,0.00401,0.00601,0.00401,0.00601,0,0.00200,0.00601,0.00401,0.00200,0,0.00200,0,0.00601,0.00401]
T=[0.13387,0.13026,0.12345,0.14028,0.13948,0.14910,0.15030,0.12625,0.16192,0.14790,0.09379,0.08617,0.10501,0.12906,0.11864,0.14269]
神经网络输入层为4，输出层为1，中间隐含层没定
遗传算法以上面神经网络所建立的模型为适度函数，每个参数的字串长度为5，4个参数组合在一起形成长度L为20的染色体，每代个体之间通过单点交叉（交叉概率为0.8）交流各自的优秀基因，及单点变异（变异概率为0.05），产生新的基因型和种群，所产生的新字串代码经解码，通过轮盘赌方法选择适应度值大的优秀基因。

回复此楼