人工智能课设项目-基于cnn-lstm的轴承故障诊断python源码(带数据+训练好的模型+详细代码注释+项目说明).zip

import torch from torch import nn class CNN(nn.Module): '''定义一维卷积神经网络模型''' def __init__(self, DEVICE, in_channel=3, out_channel=9): super(CNN, self).__init__() '''除输入层外，每个层级都包含了卷积、激活和池化三层''' '''输出层额外包含了BatchNorm层，提高网络收敛速度以及稳定性''' '''第一层卷积核大小为64，之后逐层递减''' self.layer1 = nn.Sequential( nn.Conv1d(in_channel, 16, kernel_size=64, stride=16, padding=24), nn.BatchNorm1d(16), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.layer2 = nn.Sequential( nn.Conv1d(16, 32, kernel_size=16, padding=8), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2)) self.layer3 = nn.Sequential( nn.Conv1d(32, 48, kernel_size=8, padding=4), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.layer4 = nn.Sequential( nn.Conv1d(48, 64, kernel_size=3, padding=1), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.layer5 = nn.Sequential( nn.Conv1d(64, 64, kernel_size=3), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) # 全连接层定义 引入Dropout机制以提高泛化能力 self.fc = nn.Sequential( nn.Linear(64, 128), nn.Dropout(0.3), nn.ReLU(inplace=True), nn.Linear(128, out_channel) ) # 使用softmax函数以计算输出从属于每一类的概率 self.softmax = nn.Softmax(1) def forward(self, x): '''前向传播''' x = self.layer1(x) x = self.layer2(x) x = self.layer3(x) x = self.layer4(x) x = self.layer5(x) x = x.view(x.size(0), -1) x = self.fc(x) x = x.view(x.size(0), x.size(1), 1) x = self.softmax(x) return x class LSTM_CNN(nn.Module): '''定义LSTM-CNN网络模型''' def __init__(self, DEVICE, in_channel=3, out_channel=9): super(LSTM_CNN, self).__init__() self.DEVICE = DEVICE '''LSTM相关神经元定义''' self.lstm_layer1 = nn.LSTM(in_channel, 32) self.lstm_layer2 = nn.LSTM(64, 1) self.lstm_fc1 = nn.Sequential( nn.Linear(32, 64), nn.ReLU(inplace=True), nn.Linear(64, 64) ) self.lstm_fc2 = nn.Sequential( nn.Linear(1024, 256), nn.ReLU(inplace=True), nn.Linear(256, 64) ) '''CNN相关神经元定义''' self.cnn_layer1 = nn.Sequential( nn.Conv1d(in_channel, 16, kernel_size=64, stride=16, padding=24), nn.BatchNorm1d(16), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.cnn_layer2 = nn.Sequential( nn.Conv1d(16, 32, kernel_size=3, padding=1), nn.BatchNorm1d(32), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2)) self.cnn_layer3 = nn.Sequential( nn.Conv1d(32, 64, kernel_size=3, padding=1), nn.BatchNorm1d(64), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.cnn_layer4 = nn.Sequential( nn.Conv1d(64, 64, kernel_size=3, padding=1), nn.BatchNorm1d(64), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.cnn_layer5 = nn.Sequential( nn.Conv1d(64, 64, kernel_size=3), nn.BatchNorm1d(64), nn.ReLU(inplace=True), nn.MaxPool1d(kernel_size=2, stride=2) ) self.fc = nn.Sequential( nn.Linear(128, 64), nn.Dropout(0.3), nn.ReLU(inplace=True), nn.Linear(64, out_channel) ) # 使用softmax函数以计算输出从属于每一类的概率 self.softmax = nn.Softmax(1) def forward(self, x): '''前向传播''' '''*******LSTM*******''' # 初始化隐藏神经元 x_lstm = x.permute(0, 2, 1) x_lstm.to(self.DEVICE) h1 = torch.zeros(1, 128, 32).to(self.DEVICE) c1 = torch.zeros_like(h1).to(self.DEVICE) h2 = torch.zeros(1, 128, 1).to(self.DEVICE) c2 = torch.zeros_like(h2).to(self.DEVICE) y_lstm_ = [] # 对原时序信号分段 for i in range(8): x_lstm_ = x_lstm[:, i*128:(i+1)*128] y, (h1, c1) = self.lstm_layer1(x_lstm_, (h1, c1)) y = self.lstm_fc1(y) y, (h2, c2) = self.lstm_layer2(y, (h2, c2)) y.to(self.DEVICE) y_lstm_.append(y) # 合并每一段的结果 y_lstm = torch.cat(y_lstm_, 1) y_lstm = y_lstm.view(y_lstm.size(0), -1) y_lstm = self.lstm_fc2(y_lstm) '''*******CNN*******''' x = self.cnn_layer1(x) x = self.cnn_layer2(x) x = self.cnn_layer3(x) x = self.cnn_layer4(x) x = self.cnn_layer5(x) x = x.view(x.size(0), -1) '''******LSTM+CNN******''' # 连接LSTM和CNN的输出，并通过全连接神经元 x = torch.cat([x, y_lstm], 1) x = self.fc(x) x = x.view(x.size(0), x.size(1), 1) y = self.softmax(x) return y