Python机器学习体系构建教程.zip

#!/usr/bin/env python # -*- coding: utf-8 -*- """ @Author:ChileWang @Created On 2019-05-26 @Coding Environment: Anaconda Python 3.7 """ import random import matplotlib.pyplot as plt import pandas as pd """ 问题： 从某物流中心用多台配送车辆(not limited)向多个客户送货,每个客户的位置和货物需求量一定(8t),每台配送车辆的载重量一定,其一次配送的最大行驶距离一定,要求合理安排车辆配送路线,使目标函数得到优化,并满足以下条件: (1) 每条配送路径上各客户的需求量之和不超过配送车辆的载重量;(8t) (2) 每条配送路径的长度不超过配送车辆一次配送的最大行驶距离;(80KM) (3) 每个客户的需求必须满足,且只能由一台配送车辆送货。 以配送总里程最短为目标函数 """ """ 一个实例： 某物流中心有2 台配送车辆,其载重量均为8t ,车辆每次配送的最大行驶距离为50km ,配送中心(其编号为0) 与8 个客户之间及8 个客户相互之间的距离dij 、8 个客户的货物需求量qj (i 、j = 1 ,2 , ⋯,8) 均见表1 。要求合理安排车辆配送路线,使配送总里程最短。 采用以下参数:群体规模取20 ,进化代数取25 ,交叉概率取0.9 ,变异概率取0.09 ,变异时基因换位次数取5 , 对不可行路径的惩罚权重取100km ,实施爬山操作时爬山次数取20 。对实例随机求解10 次。 """ # 遗传算法 class GeneticAlgorithm: # -----------初始数据定义--------------------- # 定义一个24 * 24的二维数组表示配送中心(编号为0)与23个客户之间，以及23个客户相互之间的距离d[i][j] d = [[0,33.3,25.5,29.7,25.1,29.1,28.7,37.3,35.3,34.3,29.6,26.8,25.6,33.6,29.3,27.2,33.4,27.9,29.1,32.8,25.9,34.4,33.3,30.4], # 配送中心（编号为0）到23个客户送货点的距离 [33.3,0.0,16.1,11.7,15.7,7.9,10.6,2.5,2.4,1.5,6.8,8.8,9.9,4.4,6.2,11.7,9.8,16.0,7.9,3.4,9.6,2.7,4.5,10.1], # 第1个客户到配送中心和其他22个客户送货点的距离 [25.5,16.1,0.0,10.8,2.6,9.8,9.3,17.9,17.3,18.2,10.2,10.3,8.5,14.4,11.0,6.4,10.5,5.3,9.7,14.4,9.1,15.0,14.0,8.0], # 第2个客户到配送中心和其他8个客户送货点的距离 [29.7,11.7,10.8,0.0,11.0,2.1,2.2,10.9,10.3,11.3,2.5,2.6,4.0,6.8,3.7,2.9,6.6,5.2,2.0,8.3,5.9,6.4,6.4,4.0], [25.1,15.7,2.6,11.0,0.0,7.4,6.3,16.3,15.6,16.6,8.6,8.6,6.6,12.8,9.4,4.5,8.9,2.8,8.1,12.8,9.7,12.5,12.3,6.8], [29.1,7.9,9.8,2.1,7.4,0.0,2.4,11.2,10.5,11.5,1.2,1.4,2.7,5.1,2.0,3.8,7.3,7.9,2.0,5.2,4.7,4.7,4.6,4.2], [28.7,10.6,9.3,2.2,6.3,2.4,0.0,10.3,9.7,10.7,10.7,4.1,5.4,5.9,2.8,3.4,5.4,6.7,1.5,6.1,9.3,5.5,5.5,2.8], [37.3,2.5,17.9,10.9,16.3,11.2,10.3,0.0,2.0,3.1,4.9,6.7,8.1,2.5,4.4,9.3,7.6,13.6,5.3,1.2,1.2,1.4,2.8,11.0], [35.3,2.4,17.3,10.3,15.6,10.5,9.7,2.0,0.0,1.5,7.2,9.1,11.3,4.7,6.7,12.7,10.4,17.0,7.7,3.5,10.2,3.7,5.0,10.7], [34.3,1.5,18.2,11.3,16.6,11.5,10.7,3.1,1.5,0.0,11.8,13.2,11.9,5.3,11.2,12.6,12.6,16.9,11.3,5.3,11.6,6.4,6.4,13.2], [29.6,6.8,10.2,2.5,8.6,1.2,10.7,4.9,7.2,11.8,0.0,1.8,3.2,5.0,1.8,4.2,7.7,8.3,1.1,5.1,5.2,4.6,4.5,4.7], [26.8,8.8,10.3,2.6,8.6,1.4,4.1,6.7,9.1,13.2,1.8,0.0,1.9,6.4,3.3,2.3,8.2,5.8,3.0,6.6,3.9,6.0,6.0,5.1], [25.6,9.9,8.5,4.0,6.6,2.7,5.4,8.1,11.3,11.9,3.2,1.9,0.0,6.9,3.7,3.3,8.2,6.4,3.5,7.0,2.8,7.7,6.4,7.4], [33.6,4.4,14.4,6.8,12.8,5.1,5.9,2.5,4.7,5.3,5.0,6.4,6.9,0.0,2.4,6.6,5.6,10.9,6.6,2.1,7.0,1.6,0.3,6.0], [29.3,6.2,11.0,3.7,9.4,2.0,2.8,4.4,6.7,11.2,1.8,3.3,3.7,2.4,0.0,6.3,5.5,10.4,2.4,3.3,5.4,2.8,2.7,6.7], [27.2,11.7,6.4,2.9,4.5,3.8,3.4,9.3,12.7,12.6,4.2,2.3,3.3,6.6,6.3,0.0,7.2,4.5,3.1,8.5,5.8,9.1,7.4,4.2], [33.4,9.8,10.5,6.6,8.9,7.3,5.4,7.6,10.4,12.6,7.7,8.2,8.2,5.6,5.5,7.2,0.0,7.5,5.2,7.4,13.7,7.0,6.9,5.1], [27.9,16.0,5.3,5.2,2.8,7.9,6.7,13.6,17.0,16.9,8.3,5.8,6.4,10.9,10.4,4.5,7.5,0.0,7.6,7.6,12.2,12.9,11.8,6.3], [29.1,7.9,9.7,2.0,8.1,2.0,1.5,5.3,7.7,11.3,1.1,3.0,3.5,6.6,2.4,3.1,5.2,7.6,0.0,5.1,6.2,4.6,4.5,3.5], [32.8,3.4,14.4,8.3,12.8,5.2,6.1,1.2,3.5,5.3,5.1,6.6,7.0,2.1,3.3,8.5,7.4,7.6,5.1,0.0,11.5,1.7,1.6,7.3], [25.9,9.6,9.1,5.9,9.7,4.7,9.3,1.2,10.2,11.6,5.2,3.9,2.8,7.0,5.4,5.8,13.7,12.2,6.2,11.5,0.0,9.1,6.5,9.7], [34.4,2.7,15.0,6.4,12.5,4.7,5.5,1.4,3.7,6.4,4.6,6.0,7.7,1.6,2.8,9.1,7.0,12.9,4.6,1.7,9.1,0.0,2.5,7.0], [33.3,4.5,14.0,6.4,12.3,4.6,5.5,2.8,5.0,6.4,4.5,6.0,6.4,0.3,2.7,7.4,6.9,11.8,4.5,1.6,6.5,2.5,0.0,5.1], [30.4,10.1,8.0,4.0,6.8,4.2,2.8,11.0,10.7,13.2,4.7,5.1,7.4,6.0,6.7,4.2,5.1,6.3,3.5,7.3,9.7,7.0,5.1,0.0], ] # 23个客户分布需要的货物的需求量，第0位为配送中心自己 q = [0, 2, 2, 2, 3, 1, 2, 1, 1, 2, 1, 1, 1, 2, 1, 3, 2, 2, 1, 1, 1, 2, 1, 3] # 定义一些遗传算法需要的参数 JCL = 0.9 # 遗传时的交叉率 BYL = 0.09 # 遗传时的变异率 JYHW = 5 # 变异时的基因换位次数 PSCS = 20 # 爬山算法时的迭代次数 def __init__(self, rows, times, mans, cars, tons, distance, PW): self.rows = rows # 排列个数 self.times = times # 迭代次数 self.mans = mans # 客户数量 self.cars = cars # 车辆总数 self.tons = tons # 车辆载重 self.distance = distance # 车辆一次行驶的最大距离 self.PW = PW # 当生成一个不可行路线时的惩罚因子 # -------------遗传函数开始执行--------------------- def run(self): print("开始迭代") # 路线数组 lines = [[0 for i in range(self.mans)] for i in range(self.rows)] # 适应度 fit = [0 for i in range(self.rows)] # 初始输入获取rows个随机排列，并且计算适应度 j = 0 for i in range(0, self.rows): j = 0 while j < self.mans: num = int(random.uniform(0, self.mans)) + 1 if not self.isHas(lines[i], num): lines[i][j] = num j += 1 # 计算每个线路的适应度 fit[i] = self.calFitness(lines[i], False) # 开始迭代 t = 0 while t < self.times: # 适应度 newLines = [[0 for i in range(self.mans)] for i in range(self.rows)] nextFit = [0 for i in range(self.rows)] randomFit = [0 for i in range(self.rows)] totalFit = 0 tmpFit = 0 # 计算总的适应度 for i in range(self.rows): totalFit += fit[i] # 通过适应度占总适应度的比例生成随机适应度 for i in range(self.rows): randomFit[i] = tmpFit + fit[i] / totalFit tmpFit += randomFit[i] # 上一代中的最优直接遗传到下一代 m = fit[0] ml = 0 for i in range(self.rows): if m < fit[i]: m = fit[i] ml = i for i in range(self.mans): newLines[0][i] = lines[ml][i] nextFit[0] = fit[ml] # 对最优解使用爬山算法促使其自我进化 self.clMountain(newLines[0]) # print "开始遗传" # 开始遗传 nl = 1 while nl < self.rows: # 根据概率选取排列 r = int(self.randomSelect(randomFit)) # 判断是否需要交叉，不能越界 if random.random() < self.JCL and nl + 1 < self.rows: fline = [0 for x in range(self.mans)] nline = [0 for x in range(self.mans)] # 获取交叉排列 rn = int(self.randomSelect(randomFit))