NP=20; #种群数量,一般取10~200,即染色体数目,将定义域划分50个(不等距)
L=6; #二进制位串长度,即染色体上基因数目,和计算要求的精度有关,如#精度10-5,则219<10*105<220,至少需要20位。‘100001’前三位’100’是x,后三位‘001’是y
Pc=0.8; #交叉率,一般取0.25~1
Pm=0.1; #变异率,一般取0.001~0.1
G=100; #最大遗传代数,一般100~1000
import random
import pandas as pd
import numpy as np
import math
#产生随机种群
def randpop(NP,L):
seed = "01"
pop2 = []
for i in range(NP):
chr = []
for j in range(L):
chr.append(random.choice(seed))
salt = ''.join(chr)
pop2.append(salt)
#print(pop2)
return pop2
#解码
def jiema(NP,pop2):
pop10 = []
for i in range(NP):
x = pop2[i][:3]
y = pop2[i][3:]
#二进制转十进制
x1 = int(x,2)
y1 = int(y,2)
#print(m)
pop10.append((x1,y1))
#print(pop10)
return pop10
#计算这个染色体对应解的适应度,即目标函数值
def getfit(NP,pop10):
fit = []
for i in range(NP):
x = pop10[i][0]
y = pop10[i][1]
z = math.pow(x,2) + math.pow(y,2)
fit.append(z)
#print(y)
#print(fit)
return fit
#轮盘赌选择
def xuanze(NP,pop2,pop10,fit):
df = pd.DataFrame({'pop2':pop2,'pop10':pop10,'fit':fit})
#print(df)
sum1 = 0
for i in range(NP):
sum1 = sum1+df.loc[i].fit
#print(sum1)
p_list = []
for i in range(NP):
p1 = df.loc[i].fit/sum1
p_list.append(p1)
#print(p_list)
#print(len(p_list))
pp_list = []
for i in range(NP):
pp = 0
for j in range(i+1):
pp1 = p_list[j]
#print(pp1)
pp = pp+pp1
#print(pp)
pp_list.append(pp)
#print(pp_list)
#print(len(pp_list))
df['p'] = p_list
df['pp'] = pp_list
#print(df)
xuanze = []
for i in range(NP):
rand = random.random()
#print(rand)
n = 0
for j in range(NP):
if rand<pp_list[j]:
n = j
break
#print(n)
m = df.loc[n].pop2
#print(m)
xuanze.append(m)
#print(xuanze)
df['new'] = xuanze
#print(df)
#返回新的01字符串
return xuanze
#执行基于概率Pc的两点交叉操作,采用奇偶交叉,即第1个和第2个交叉,第3个和第4个交叉,依次进行
def jiaocha(NP,Pc,L,pop2):
pop2_ji = pop2[::2]
pop2_ou = pop2[1::2]
#print(pop2_ji)
#print(pop2_ou)
for i in range(int(NP/2)):
rand = random.random()
#print(rand)
if rand<Pc: #控制交叉的染色体总数
r = random.randint(0,L-1)
#print(r) #随机生成要交叉的基因位置
#复制一个新的再交换
pji = pop2_ji[i][r:]
pou = pop2_ou[i][r:]
pop2_ji[i] = pop2_ji[i][0:r] + pou
#print(pop2_ji[i])
pop2_ou[i] = pop2_ou[i][0:r] + pji
#print(pop2_ou[i])
#print(pop2_ji)
#print(pop2_ou)
pop2_new = pop2_ji + pop2_ou
#print(pop2_new)
return pop2_new
#执行基于概率Pm的变异操作,即随机选择一个基因位置取反,若是0则变为1,若是1则变为0
def bianyi(NP,Pm,L,pop2_new):
#print(pop2_new)
for i in range(NP):
rand = random.random()
#print(rand)
if rand<Pm:
r = random.randint(0,L-1)
#print(r) #随机生成要变异的基因位置
if pop2_new[i][r]=='0':
p = list(pop2_new[i])
p[0]='1'
pop2_new[i] = "".join(p)
else:
p = list(pop2_new[i])
p[0]='0'
pop2_new[i] = "".join(p)
#print(pop2_new)
return pop2_new
if __name__ == "__main__":
NP=20; #种群数量,一般取10~200,即染色体数目,将定义域划分50个(不等距)
L=6; #二进制位串长度,即染色体上基因数目,和计算要求的精度有关,如
#精度10^-5,则2^19<10*10^5<2^20,至少需要20位
Pc=0.8; #交叉率,一般取0.25~1
Pm=0.1; #变异率,一般取0.001~0.1
G=100; #最大遗传代数,一般100~1000
trace = []
#随机产生种群01字符串
pop2 = randpop(NP,L)
#print("pop2:",pop2)
for i in range(G):
#解码
pop10 = jiema(NP,pop2)
#print("pop10:",pop10)
#求适应度
fit = getfit(NP,pop10)
#print("fit:",fit)
print("第"+str(i)+"次:")
df = pd.DataFrame({'pop2':pop2,'pop10':pop10,'fit':fit})
print(df)
#选择
pop2 = xuanze(NP,pop2,pop10,fit)
#print("pop2:",pop2)
#交叉
pop2_new = jiaocha(NP,Pc,L,pop2)
#变异
pop2 = bianyi(NP,Pm,L,pop2_new)
#print("pop2_new:",pop2)