此次使用python实现蚁群算法是仿照蚁群优化算法的JAVA实现中的蚁群算法实现方法，使用的也是其中的数据(此处为上传数据)，如需更深一步了解蚁群算法原理和具体实现过程，请参考蚁群优化算法的JAVA实现和蚁群算法java实现以及TSP问题蚁群算法求解。

# ants 类

import numpy as np

import random

import math

class ant:

Tour = 0

unvisitedcity = 0

tourlength = 0

citys = 0

count = 0

def RandomSelectCity(self,citycount):

ant.citys = citycount

ant.tourlength = 0

ant.Tour = np.zeros((citycount+1),int)

ant.unvisitedcity = np.ones((citycount))

for i in np.arange(citycount):

ant.Tour[i] = -1

firstcity = random.randint(0,citycount-1)

ant.unvisitedcity[firstcity] = 0 #0表示已经访问过

ant.Tour[0] = firstcity #起始城市

def SelectNextCity(self,index,tao,distance):

p = np.zeros((ant.citys)) #citys

x = 0

alpha = 1.0

beta = 2.0

sum = 0

currentcity = ant.Tour[index-1]

# print("curr",ant.Tour)

# print(ant.unvisitedcity)

for i in range(ant.citys):

# print(ant.unvisitedcity[i])

if ant.unvisitedcity[i] == 1:

sum += (math.pow(tao[currentcity][i], alpha) * math.pow(1.0/distance[currentcity][i], beta))

# 计算每个城市被选中的概率

for i in range(ant.citys):

if ant.unvisitedcity[i] == 0:

p[i] = 0.0

else:

p[i] = (math.pow(tao[currentcity][i], alpha) * math.pow(1.0/distance[currentcity][i], beta))/sum

selectp = random.random()

# 轮盘赌选择一个城市

sumselect = 0.0

selectcity = -1

for i in range(ant.citys):

sumselect += p[i]

# print("i",i,"sumselect",sumselect)

if sumselect >= selectp:

selectcity = i

break

if selectcity == -1:

print('')

ant.Tour[index] = selectcity

# print("selectcity",selectcity)

ant.unvisitedcity[selectcity] = 0

# print(ant.Tour)

# 计算蚂蚁获得的路径的长度

# distance 全局的距离矩阵信息

def CalTourLength(self,distance):

ant.Tour[ant.citys] = ant.Tour[0]

for i in range(ant.citys):

ant.tourlength += distance[int(ant.Tour[i])][int(ant.Tour[i+1])]

# ACO类

from ACO_implement.ACO_implement import ants

import numpy as np

import math

import sys

class ACO:

# 定义蚂蚁群

ants = []

antcount = 0 # 蚂蚁数量

distance = 0 #城市之间的距离

tao = 0 #信息素矩阵

citycount = 0 #城市数量

besttour = 0 #求解的最佳路径

bestlength = 0 #求解最优解的长度

def __init__(self,filename,antnum):

ACO.antcount = antnum

#读取数据

x = []

y = []

z = []

str = open(filename).readlines()

ACO.citycount = int(str[0])

for i in range(1,ACO.citycount+1):

strbuff = str[i].split()

x.append(int(strbuff[1]))

y.append(int(strbuff[2]))

# 计算距离矩阵，计算的是点之间的距离

ACO.distance = np.zeros((ACO.citycount,ACO.citycount))

for i in range(ACO.citycount):

for j in range(ACO.citycount):

ACO.distance[i][j] = math.sqrt(pow(x[i]-x[j],2)+pow(y[i]-y[j],2))

# 初始化信息矩阵

ACO.tao = np.zeros((ACO.citycount,ACO.citycount))

for i in range(ACO.citycount):

for j in range(ACO.citycount):

ACO.tao[i][j] = 0.1

ACO.bestlength = sys.maxsize #最优路径长度

ACO.besttour = np.zeros((ACO.citycount)) #最优路径

# 随机放置蚂蚁

# for i in range(ACO.antcount):

# ACO.ants.append(ants.ant())

# ACO.ants[i].RandomSelectCity(ACO.citycount)

"""

跟新信息素矩阵

"""

def UpdateTao(self):

rou = 0.5

# 信息素挥发

for i in range(ACO.citycount):

for j in range(ACO.citycount):

ACO.tao[i][j] = ACO.tao[i][j] *(1-rou)

# 信息素更新

for i in range(ACO.citycount):

for j in range(ACO.citycount):

ACO.tao[int(ACO.ants[i].Tour[j])][ACO.ants[i].Tour[j+1]] += 1.0/ACO.ants[i].tourlength

"""

ACO的运行过程

maxgen ACO的最多次循环次数

"""

def run(self,maxgen):

# 每只蚂蚁移动的过程

for runtimes in range(maxgen):

# 随机放置蚂蚁

for i in range(ACO.antcount):

ACO.ants.append(ants.ant())

ACO.ants[i].RandomSelectCity(ACO.citycount)

for j in range(1,ACO.citycount):

ACO.ants[i].SelectNextCity(j,ACO.tao,ACO.distance)

# 计算蚂蚁获得的路径的长度

ACO.ants[i].CalTourLength(ACO.distance)

if ACO.ants[i].tourlength < ACO.bestlength:

# 保留最优路径

ACO.bestlength = ACO.ants[i].tourlength

print("第",runtimes+1,"次,发现解",ACO.bestlength)

for j in range(0,ACO.citycount):

ACO.besttour[j] = ACO.ants[i].Tour[j]

ACO.UpdateTao(self)

for i in range(ACO.antcount):

ACO.ants[i].RandomSelectCity(ACO.citycount)

"""输出运行结果"""

def ReportResult(self):

print("最优路径长度是",ACO.bestlength)

print("最优路径为：",ACO.besttour+1)

#main类

from ACO_implement.ACO_implement import ACO

aco = ACO.ACO("ant_data",1000)

# aco.init("ant_data",1000)

#点数量为50，蚂蚁数量

aco.run(200)

aco.ReportResult()