五维的电力负荷数据

好的，为了比较 LSTM 和 Transformer 在电力负荷预测中的性能，我将为您提供一个示例代码。假设您的电力负荷数据包含以下五个维度：年份、月份、日期、小时和负荷值。数据的格式可以是 CSV 或 Excel 文件。

在下面的代码中，我们首先导入必要的库，然后读取数据文件。接下来，我们将数据划分为训练集和测试集，并进行归一化处理。然后，我们使用 Keras 库构建 LSTM 和 Transformer 模型，并将它们训练在训练集上。最后，我们使用测试集来评估模型的性能，并将结果可视化。

import pandas as pd
import numpy as np
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential
from keras.layers import Dense, LSTM
from keras.layers import TransformerBlock, MultiHeadAttention, Input, Dropout
import matplotlib.pyplot as plt

# 读取数据
data = pd.read_csv('electricity.csv')

# 将数据转换为 Numpy 数组
data = np.array(data)

# 划分训练集和测试集
train_size = int(len(data) * 0.8)
test_size = len(data) - train_size
train_data, test_data = data[0:train_size,:], data[train_size:len(data),:]

# 归一化
scaler = MinMaxScaler(feature_range=(0, 1))
train_data = scaler.fit_transform(train_data)
test_data = scaler.transform(test_data)

# 准备数据
def prepare_data(data, look_back):
    X, Y = [], []
    for i in range(len(data)-look_back-1):
        X.append(data[i:(i+look_back), :])
        Y.append(data[(i+look_back), 4])
    return np.array(X), np.array(Y)

look_back = 24*7
train_X, train_Y = prepare_data(train_data, look_back)
test_X, test_Y = prepare_data(test_data, look_back)

# 构建 LSTM 模型
model_lstm = Sequential()
model_lstm.add(LSTM(50, input_shape=(train_X.shape[1], train_X.shape[2])))
model_lstm.add(Dense(1))
model_lstm.compile(loss='mse', optimizer='adam')

# 训练 LSTM 模型
history_lstm = model_lstm.fit(train_X, train_Y, epochs=50, batch_size=64, validation_data=(test_X, test_Y), verbose=2, shuffle=False)

# 构建 Transformer 模型
def build_transformer(n_steps, n_features, d_model, n_heads, d_ff, dropout):
    input_layer = Input(shape=(n_steps, n_features))
    transformer_block = TransformerBlock(d_model, n_heads, d_ff, dropout)
    attention_layer = MultiHeadAttention(n_heads=n_heads, key_dim=d_model // n_heads)
    x = transformer_block(input_layer)
    x = attention_layer(x, x)
    x = Dropout(dropout)(x)
    output_layer = Dense(1)(x)
    model = keras.models.Model(inputs=input_layer, outputs=output_layer)
    model.compile(loss='mse', optimizer='adam')
    return model

model_transformer = build_transformer(train_X.shape[1], train_X.shape[2], d_model=256, n_heads=8, d_ff=1024, dropout=0.1)

# 训练 Transformer 模型
history_transformer = model_transformer.fit(train_X, train_Y, epochs=50, batch_size=64, validation_data=(test_X, test_Y), verbose=2, shuffle=False)

# 评估模型
train_predict_lstm = model_lstm.predict(train_X)
test_predict_lstm = model_lstm.predict(test_X)

train_predict_transformer = model_transformer.predict(train_X)
test_predict_transformer = model_transformer.predict(test_X)

train_predict_lstm = scaler.inverse_transform(train_predict_lstm)
train_Y_lstm = scaler.inverse_transform([train_Y])

test_predict_lstm = scaler.inverse_transform(test_predict_lstm)
test_Y_lstm = scaler.inverse_transform([test_Y])

train_predict_transformer = scaler.inverse_transform(train_predict_transformer)
train_Y_transformer = scaler.inverse_transform([train_Y])

test_predict_transformer = scaler.inverse_transform(test_predict_transformer)
test_Y_transformer = scaler.inverse_transform([test_Y])

# 可视化结果
plt.plot(train_Y_lstm[0], label='Actual')
plt.plot(train_predict_lstm[:,0], label='LSTM')
plt.plot(train_predict_transformer[:,0], label='Transformer')
plt.legend()
plt.show()

plt.plot(test_Y_lstm[0], label='Actual')
plt.plot(test_predict_lstm[:,0], label='LSTM')
plt.plot(test_predict_transformer[:,0], label='Transformer')
plt.legend()
plt.show()

这是一个基本的示例代码，您可以根据您的具体情况进行修改和调整。请确保您已安装并导入了所需的库。