python开发基于深度强化学习(DQN)解决三维在线装箱问题源码+项目说明文档+模型.zip

# -*- coding: utf-8 -*- import torch import torch.nn as nn from torch.autograd import Variable import torch.nn.functional as F import numpy as np import random import copy import time # import draw from container import * from data import * # from draw import * # 1. Define some Hyper Parameters EPSILON = 0.9 # epsilon used for epsilon greedy approach BATCH_SIZE = 16 # batch size of sampling process from buffer LR = 0.0001 # learning rate GAMMA = 0.9 # discount factor TARGET_NETWORK_REPLACE_FREQ = 100 # How frequently target netowrk updates MEMORY_CAPACITY = 2000 # The capacity of experience replay buffer device=torch.device('cuda' if torch.cuda.is_available() else 'cpu') # 2. Random generate box data solution = [[(91, 54, 45, 32), (105, 77, 72, 24), (79, 78, 48, 30)], [(108, 76, 30, 24), (110, 43, 25, 7), (92, 81, 55, 22), (81, 33, 28, 13), (120, 99, 73, 15)], [(88, 54, 39, 16), (94, 54, 36, 14), (87, 77, 43, 20), (100, 80, 72, 16), (83, 40, 36, 6),(91, 54, 22, 15), (109, 58, 54, 17), (94, 55, 30, 9)], [(86, 84, 45, 18), (81, 45, 34, 19), (70, 54, 37, 13), (71, 61, 52, 16), (78, 73, 40, 10),(69, 63, 46, 13), (72, 67, 56, 10), (75, 75, 36, 8), (94, 88, 50, 12), (65, 51, 50, 13)], [(108, 76, 30, 12), (110, 43, 25, 12), (92, 81, 55, 6), (81, 33, 28, 9), (120, 99, 73, 5), (111, 70, 48, 12), (98, 72, 46, 9), (95, 66, 31, 10), (85, 84, 30, 8), (71, 32, 25, 3), (36, 34, 25, 10), (97, 67, 62, 7), (33, 25, 23, 7), (95, 27, 26, 10), (94, 81, 44, 9)]] def random_generate(): idx = np.random.randint(0,len(solution)) box_list = solution[idx] gen_box_order = [] while True: index = np.random.randint(0, len(box_list)) box_list[index] = (box_list[index][0], box_list[index][1], box_list[index][2], box_list[index][3]-1) gen_box_order.append((box_list[index][0], box_list[index][1], box_list[index][2])) if box_list[index][3] == 0: box_list.pop(index) if len(box_list) == 0: break return gen_box_order def normalization(data): _range = np.max(data) - np.min(data) return (data - np.min(data)) / _range def standardization(data): mu = np.mean(data) sigma = np.std(data) return (data - mu) / sigma # 3. Define the network used in both target net and the net for training class CNN(nn.Module): def __init__(self): super(CNN, self).__init__() # 继承__init__功能 ## 第一层卷积 self.conv1 = nn.Sequential( # 输入[2,587,233] nn.Conv2d( in_channels=2, # 输入图片的高度 out_channels=16, # 输出图片的高度 kernel_size=3, # 5x5的卷积核，相当于过滤器 stride=1, # 卷积核在图上滑动，每隔一个扫一次 padding=1, # 给图外边补上0 ), # 经过卷积层 输出[16,28,28] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[16,14,14] 传入下一个卷积 ) ## 第二层卷积 self.conv2 = nn.Sequential( nn.Conv2d( in_channels=16, # 同上 out_channels=32, kernel_size=3, stride=1, padding=1 ), # 经过卷积 输出[32, 14, 14] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[32,7,7] 传入输出层 ) ## 第三层卷积 self.conv3 = nn.Sequential( nn.Conv2d( in_channels=32, # 同上 out_channels=64, kernel_size=3, stride=1, padding=1 ), # 经过卷积 输出[32, 14, 14] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[32,7,7] 传入输出层 ) ## 第四层卷积 self.conv4 = nn.Sequential( nn.Conv2d( in_channels=64, # 同上 out_channels=128, kernel_size=3, stride=1, padding=1 ), # 经过卷积 输出[32, 14, 14] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[32,7,7] 传入输出层 ) ## 第五层卷积 self.conv5 = nn.Sequential( nn.Conv2d( in_channels=128, # 同上 out_channels=256, kernel_size=3, stride=1, padding=1 ), # 经过卷积 输出[32, 14, 14] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[32,7,7] 传入输出层 ) ## 第六层卷积 self.conv6 = nn.Sequential( nn.Conv2d( in_channels=256, # 同上 out_channels=512, kernel_size=3, stride=1, padding=1 ), # 经过卷积 输出[32, 14, 14] 传入池化层 nn.ReLU(), nn.MaxPool2d(kernel_size=2) # 经过池化 输出[32,7,7] 传入输出层 ) ## 输出层 self.output = nn.Linear(in_features=512*9*3, out_features=1) def forward(self, x): x = self.conv1(x) x = self.conv2(x) # [batch, 32,7,7] x = self.conv3(x) # [batch, 32,7,7] x = self.conv4(x) # [batch, 32,7,7] x = self.conv5(x) # [batch, 32,7,7] x = self.conv6(x) # [batch, 32,7,7] x = x.view(x.size(0), -1) # 保留batch, 将后面的乘到一起 [batch, 32*7*7] output = self.output(x) # 输出[50,10] return output class DQN(object): def __init__(self): # -----------Define 2 networks (target and training)------# self.eval_net, self.target_net = CNN(), CNN() # Define counter, memory size and loss function self.learn_step_counter = 0 # count the steps of learning process self.memory: List = [None] * MEMORY_CAPACITY self.memory_counter = 0 # counter used for experience replay buffer # ------- Define the optimizer------# self.optimizer = torch.optim.Adam(self.eval_net.parameters(), lr=LR) # ------Define the loss function-----# self.loss_func = nn.MSELoss() def choose_action(self, state:Container, cargo): # 可行点取最大的 is_encase, inputs, points, poses = state.encase(cargo) # torch.set_printoptions(profile="full") # print(inputs) if is_encase == False: return is_encase, is_encase, is_encase if np.random.uniform() < EPSILON: # greedy data = inputs.data.cpu().numpy() data = normalization(data) data = standardization(data) inputs = torch.tensor(data) with torch.no_grad(): actions_value = self.target_net.forward(inputs) action = torch.max(actions_value, 0)[1].data.numpy() action = action[0] point = points[action] pose = poses[action] else: action = np.random.randint(0, high=len(points)) point = points[action] pose = poses[action] return point, pose def store_transition(self, s:torch.Tensor, a:Cargo, r, s_:Container, a_:Cargo): transition = [s, a.matrix(), r, s_, a_] # if the capacity is full, then use index to replace the old memory with new one index = self.memory_counter % MEMORY_CAPACITY self.memory[index] = transition self.memory_counter += 1 def learn(self): # update the target network every fixed steps if self.learn_step_counter % TARGET_NETWORK_REPLACE_FREQ == 0: # Assign the parameters of eval_net to target_net self.target_net.load_sta