基于pytorch的LSTM多变量多输出时间序列预测使用例

import numpy as np import torch import torch.nn as nn device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") x=np.arange(0,150.1,0.1).tolist() #print(x.shape) sin_y=np.sin(x) cos_y=np.cos(x) tan_y=np.tan(x) y_set=np.array([sin_y,cos_y,cos_y]) #print(y_set[:,0]) x_test=[] y_test=[] for i in range(1000): ''' 建立训练集，x为sin_y、cos_y、tan_y构成的(50,3)的数组，y为下一个（比如x为i列到i+50列，y就是第i+51列）(1，3)的数组 ''' x_test.append(y_set[:,i:i+50].T.tolist()) y_test.append(y_set[:,i+51].tolist()) print(np.array(x_test)) print(np.array(x_test).shape) print(np.array(y_test).shape) np.random.seed(168) np.random.shuffle(x_test) np.random.seed(168) np.random.shuffle(y_test) x=torch.tensor(x_test).to(device) y=torch.tensor(y_test).to(device) epoch=10#训练10代 input_size=3#词向量的维度，即是(50,3)中的3 hidden_size=16#隐藏层维度 num_layers=7#层数 output_size=3#输出的维度，即是(1,3)中的3 batch_size=32#一次训练的样本数 class MyDataset(): def __init__(self, data): self.data = data def __getitem__(self, item): return self.data[item] def __len__(self): return len(self.data) class LSTM(nn.Module): def __init__(self, ): super().__init__() self.input_size = input_size self.hidden_size = hidden_size self.num_layers = num_layers self.output_size = output_size self.num_directions = 1 # 单向LSTM self.batch_size = batch_size self.lstm = nn.LSTM(self.input_size, self.hidden_size, self.num_layers, batch_first=True) self.linear = nn.Linear(self.hidden_size, self.output_size) def forward(self, input_seq): h_0 = torch.randn(self.num_directions * self.num_layers, self.batch_size, self.hidden_size).to(device) c_0 = torch.randn(self.num_directions * self.num_layers, self.batch_size, self.hidden_size).to(device) # output(batch_size, seq_len, num_directions * hidden_size) output, _ = self.lstm(input_seq, (h_0, c_0)) # output(5, 30, 64) pred = self.linear(output) # (5, 30, 1) #print(pred) pred = pred[:, -1, :] # (5, 1) return pred net=LSTM() net = torch.load('model_LSTM_test.pth') net = net.to(device) criterion=nn.MSELoss()#均方误差计算损失率 optimizer=torch.optim.Adam(net.parameters(),lr=1e-2) n=0 for i in range(epoch): for j in range(int(len(y)/batch_size)): train_X=x[j*batch_size:(j+1)*batch_size,:,:]#切分一个batch train_Y=y[j*batch_size:(j+1)*batch_size,:] var_x=torch.tensor(train_X,dtype=torch.float32)#,device=device var_y=torch.tensor(train_Y,dtype=torch.float32)#,device=device out = net(var_x)#输出 optimizer.zero_grad() loss=criterion(out,var_y) loss.backward() optimizer.step() if i%1==0: print('Epoch: {:4}, Loss: {:.5f}'.format(i, loss.item())) #torch.save(net, 'model_LSTM_test.pth')